From fb4279293472b408eaa7ee403927a2717e8aa991 Mon Sep 17 00:00:00 2001 From: OliverB Date: Thu, 13 Feb 2020 10:10:36 +0100 Subject: [PATCH 01/63] Update README.md --- README.md | 6 ------ 1 file changed, 6 deletions(-) diff --git a/README.md b/README.md index ae42214..4cd22d1 100644 --- a/README.md +++ b/README.md @@ -62,12 +62,6 @@ Things I will work on next: **[ ]** Approximate Nearest Neighbor Search for SentenceVectors -**[ ]** Discrete Cosine Transform Embedding (?) - -**[ ]** VLAVE Embedding (?) - -**[ ]** PowerMeans Embedding (?) - Installation From fe9af44ec887fad92ee05b07273907b154889957 Mon Sep 17 00:00:00 2001 From: OliverB Date: Thu, 13 Feb 2020 18:47:52 +0100 Subject: [PATCH 02/63] Code style fix --- README.md | 1 + fse/models/utils.py | 3 ++- 2 files changed, 3 insertions(+), 1 deletion(-) diff --git a/README.md b/README.md index 4cd22d1..aff4a04 100644 --- a/README.md +++ b/README.md @@ -2,6 +2,7 @@ Build Status Coverage Status Downloads +Language grade: Python Code style: black

diff --git a/fse/models/utils.py b/fse/models/utils.py index 4567546..190f97f 100644 --- a/fse/models/utils.py +++ b/fse/models/utils.py @@ -76,9 +76,10 @@ def compute_principal_components( current_mem = INF sample_size = len(vectors) - while 1: + while current_mem >= cache_size_gb: current_mem = sample_size * vectors.shape[1] * dtype(REAL).itemsize / 1024 ** 3 if current_mem < cache_size_gb: + # Skip if full dataset is already smaller than cache_size_gb break sample_size *= 0.995 sample_size = int(sample_size) From db4c0213aa423559847b9341679db17ad201e050 Mon Sep 17 00:00:00 2001 From: OliverB Date: Fri, 14 Feb 2020 16:40:29 +0100 Subject: [PATCH 03/63] Added file closing op --- fse/test/test_average.py | 3 ++- fse/test/test_base_s2v.py | 3 ++- fse/test/test_sentencevectors.py | 3 ++- fse/test/test_sif.py | 3 ++- fse/test/test_usif.py | 3 ++- 5 files changed, 10 insertions(+), 5 deletions(-) diff --git a/fse/test/test_average.py b/fse/test/test_average.py index 35e8e9a..983ab86 100644 --- a/fse/test/test_average.py +++ b/fse/test/test_average.py @@ -25,7 +25,8 @@ CORPUS = Path("fse/test/test_data/test_sentences.txt") DIM = 5 W2V = Word2Vec(min_count=1, size=DIM) -SENTENCES = [l.split() for i, l in enumerate(open(CORPUS, "r"))] +with open(CORPUS, "r") as f: + SENTENCES = [l.split() for i, l in enumerate(f)] W2V.build_vocab(SENTENCES) W2V.wv.vectors[:,] = np.arange(len(W2V.wv.vectors), dtype=np.float32)[:, None] diff --git a/fse/test/test_base_s2v.py b/fse/test/test_base_s2v.py index d9265ff..4193b30 100644 --- a/fse/test/test_base_s2v.py +++ b/fse/test/test_base_s2v.py @@ -27,7 +27,8 @@ CORPUS = Path("fse/test/test_data/test_sentences.txt") DIM = 5 W2V = Word2Vec(min_count=1, size=DIM) -SENTENCES = [l.split() for i, l in enumerate(open(CORPUS, "r"))] +with open(CORPUS, "r") as f: + SENTENCES = [l.split() for i, l in enumerate(f)] W2V.build_vocab(SENTENCES) diff --git a/fse/test/test_sentencevectors.py b/fse/test/test_sentencevectors.py index 00b7ba4..1fb18ec 100644 --- a/fse/test/test_sentencevectors.py +++ b/fse/test/test_sentencevectors.py @@ -26,7 +26,8 @@ CORPUS = Path("fse/test/test_data/test_sentences.txt") DIM = 5 W2V = Word2Vec(min_count=1, size=DIM, seed=42) -SENTENCES = [l.split() for l in open(CORPUS, "r")] +with open(CORPUS, "r") as f: + SENTENCES = [l.split() for i, l in enumerate(f)] W2V.build_vocab(SENTENCES) np.random.seed(42) W2V.wv.vectors = np.random.uniform(size=W2V.wv.vectors.shape).astype(np.float32) diff --git a/fse/test/test_sif.py b/fse/test/test_sif.py index a64f13f..d784b64 100644 --- a/fse/test/test_sif.py +++ b/fse/test/test_sif.py @@ -19,7 +19,8 @@ CORPUS = Path("fse/test/test_data/test_sentences.txt") DIM = 50 W2V = Word2Vec(min_count=1, size=DIM) -SENTENCES = [l.split() for l in open(CORPUS, "r")] +with open(CORPUS, "r") as f: + SENTENCES = [l.split() for i, l in enumerate(f)] W2V.build_vocab(SENTENCES) diff --git a/fse/test/test_usif.py b/fse/test/test_usif.py index e188b60..fe5320f 100644 --- a/fse/test/test_usif.py +++ b/fse/test/test_usif.py @@ -15,7 +15,8 @@ CORPUS = Path("fse/test/test_data/test_sentences.txt") DIM = 50 W2V = Word2Vec(min_count=1, size=DIM) -SENTENCES = [l.split() for l in open(CORPUS, "r")] +with open(CORPUS, "r") as f: + SENTENCES = [l.split() for i, l in enumerate(f)] W2V.build_vocab(SENTENCES) From 296621a81645b9e063d0abcfb37c7b0a5ea0d326 Mon Sep 17 00:00:00 2001 From: OliverB Date: Fri, 14 Feb 2020 18:23:29 +0100 Subject: [PATCH 04/63] Fixed lgtm test exclusion --- .lgtm.yml | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/.lgtm.yml b/.lgtm.yml index a1b5814..9f206b4 100644 --- a/.lgtm.yml +++ b/.lgtm.yml @@ -1,6 +1,6 @@ path_classifiers: test: - - exclude: "**/test_*" + - test extraction: python: From 48333b8d59ed02b24d2f0b5c181b36f026fe1fc7 Mon Sep 17 00:00:00 2001 From: OliverB Date: Fri, 14 Feb 2020 18:25:19 +0100 Subject: [PATCH 05/63] Fixed call to child in super __init__ --- fse/inputs.py | 18 +++++++++--------- fse/test/test_inputs.py | 24 ++++++++++++++++-------- 2 files changed, 25 insertions(+), 17 deletions(-) diff --git a/fse/inputs.py b/fse/inputs.py index 5671c0e..d69d397 100644 --- a/fse/inputs.py +++ b/fse/inputs.py @@ -150,6 +150,9 @@ def __init__(self, *args:[list, set, ndarray], custom_index:[list, ndarray]): """ self.custom_index = custom_index + if len(args) > 1: + RuntimeError("Argument merging not supported") + super(CIndexedList, self).__init__(*args) if len(self.items) != len(self.custom_index): @@ -176,9 +179,6 @@ def insert(self, i:int, item:str): def append(self, item:str): raise NotImplementedError("Method currently not supported") - - def extend(self, arg:[list, set, ndarray]): - raise NotImplementedError("Method currently not supported") class SplitIndexedList(BaseIndexedList): @@ -220,6 +220,9 @@ def __init__(self, *args:[list, set, ndarray], custom_index:[list, ndarray]): """ self.custom_index = custom_index + if len(args) > 1: + RuntimeError("Argument merging not supported") + super(SplitCIndexedList, self).__init__(*args) if len(self.items) != len(self.custom_index): @@ -248,9 +251,6 @@ def insert(self, i:int, item:str): def append(self, item:str): raise NotImplementedError("Method currently not supported") - def extend(self, arg:[list, set, ndarray]): - raise NotImplementedError("Method currently not supported") - class CSplitIndexedList(BaseIndexedList): def __init__(self, *args:[list, set, ndarray], custom_split:callable): @@ -296,6 +296,9 @@ def __init__(self, *args:[list, set, ndarray], custom_split:callable, custom_ind """ self.custom_split = custom_split self.custom_index = custom_index + + if len(args) > 1: + RuntimeError("Argument merging not supported") super(CSplitCIndexedList, self).__init__(*args) @@ -323,9 +326,6 @@ def insert(self, i:int, item:str): def append(self, item:str): raise NotImplementedError("Method currently not supported") - - def extend(self, arg:[list, set, ndarray]): - raise NotImplementedError("Method currently not supported") class IndexedLineDocument(object): diff --git a/fse/test/test_inputs.py b/fse/test/test_inputs.py index 259806c..97bef70 100644 --- a/fse/test/test_inputs.py +++ b/fse/test/test_inputs.py @@ -134,8 +134,10 @@ def test_mutable_funcs(self): self.il.insert(0, "the") with self.assertRaises(NotImplementedError): self.il.append("the") - with self.assertRaises(NotImplementedError): - self.il.extend(["the", "dog"]) + + def test_arg_merging(self): + with self.assertRaises(RuntimeError): + CIndexedList(self.list_a, self.list_a, custom_index=[1, 1]) class TestCSplitIndexedList(unittest.TestCase): @@ -168,8 +170,10 @@ def test_mutable_funcs(self): self.il.insert(0, "the") with self.assertRaises(NotImplementedError): self.il.append("the") - with self.assertRaises(NotImplementedError): - self.il.extend(["the", "dog"]) + + def test_arg_merging(self): + with self.assertRaises(RuntimeError): + SplitCIndexedList(self.list_a, self.list_a, custom_index=[1, 1]) class TestCSplitCIndexedList(unittest.TestCase): @@ -195,9 +199,13 @@ def test_mutable_funcs(self): self.il.insert(0, "the") with self.assertRaises(NotImplementedError): self.il.append("the") - with self.assertRaises(NotImplementedError): - self.il.extend(["the", "dog"]) - + + def test_arg_merging(self): + with self.assertRaises(RuntimeError): + CSplitCIndexedList(self.list_a, self.list_a, + custom_split=self.split_func, + custom_index=[1, 1] + ) class TestIndexedLineDocument(unittest.TestCase): def setUp(self): @@ -225,4 +233,4 @@ def test_yield(self): logging.basicConfig( format="%(asctime)s : %(levelname)s : %(message)s", level=logging.DEBUG ) - unittest.main() + unittest.main() \ No newline at end of file From 8c21256447170e33a840b1bd414a3f52c9653d38 Mon Sep 17 00:00:00 2001 From: OliverB Date: Sun, 16 Feb 2020 10:31:54 +0100 Subject: [PATCH 06/63] Added Pooling model --- fse/models/pooling.py | 351 +++++++++++++++++++++++++++++++++++++++ fse/test/test_pooling.py | 326 ++++++++++++++++++++++++++++++++++++ 2 files changed, 677 insertions(+) create mode 100644 fse/models/pooling.py create mode 100644 fse/test/test_pooling.py diff --git a/fse/models/pooling.py b/fse/models/pooling.py new file mode 100644 index 0000000..2f5c009 --- /dev/null +++ b/fse/models/pooling.py @@ -0,0 +1,351 @@ +#!/usr/bin/env python +# -*- coding: utf-8 -*- + +# Author: Oliver Borchers +# Copyright (C) 2019 Oliver Borchers + +"""This module implements the base class to compute Max Pooling representations for sentences, using highly optimized C routines, +data streaming and Pythonic interfaces. + +The implementation is based on Shen et al. (2018): Baseline Needs More Love: On Simple Word-Embedding-Based Models and Associated Pooling Mechanisms. +For more information, see . + +The training algorithms is based on the Gensim implementation of Word2Vec, FastText, and Doc2Vec. +For more information, see: :class:`~gensim.models.word2vec.Word2Vec`, :class:`~gensim.models.fasttext.FastText`, or +:class:`~gensim.models.doc2vec.Doc2Vec`. + +Initialize and train a :class:`~fse.models.pooling.MaxPooling` model + +.. sourcecode:: pycon + + >>> from gensim.models.word2vec import Word2Vec + >>> sentences = [["cat", "say", "meow"], ["dog", "say", "woof"]] + >>> model = Word2Vec(sentences, min_count=1, size=20) + + >>> from fse.models.pooling import MaxPooling + >>> avg = MaxPooling(model) + >>> avg.train([(s, i) for i, s in enumerate(sentences)]) + >>> avg.sv.vectors.shape + (2, 20) + +""" + +from __future__ import division + +from fse.models.base_s2v import BaseSentence2VecModel + +from gensim.models.keyedvectors import BaseKeyedVectors +from gensim.models.utils_any2vec import ft_ngram_hashes + +from numpy import ( + ndarray, + float32 as REAL, + sum as np_sum, + multiply as np_mult, + zeros, + amax as np_amax, + maximum as np_maximum, +) + +from typing import List + +import logging + +logger = logging.getLogger(__name__) + + +def train_pooling_np( + model: BaseSentence2VecModel, + indexed_sentences: List[tuple], + target: ndarray, + memory: ndarray, +) -> [int, int]: + """Training on a sequence of sentences and update the target ndarray. + + Called internally from :meth:`~fse.models.average.Average._do_train_job`. + + Warnings + -------- + This is the non-optimized, pure Python version. If you have a C compiler, + fse will use an optimized code path from :mod:`fse.models.average_inner` instead. + + Parameters + ---------- + model : :class:`~fse.models.base_s2v.BaseSentence2VecModel` + The BaseSentence2VecModel model instance. + indexed_sentences : iterable of tuple + The sentences used to train the model. + target : ndarray + The target ndarray. We use the index from indexed_sentences + to write into the corresponding row of target. + memory : ndarray + Private memory for each working thread + + Returns + ------- + int, int + Number of effective sentences (non-zero) and effective words in the vocabulary used + during training the sentence embedding. + + """ + size = model.wv.vector_size + vocab = model.wv.vocab + + w_vectors = model.wv.vectors + w_weights = model.word_weights + + s_vectors = target + + is_ft = model.is_ft + + mem = memory[0] + + hierarchical = model.hierarchical + window = model.window_size + + if is_ft: + # NOTE: For Fasttext: Use wv.vectors_vocab + # Using the wv.vectors from fasttext had horrible effects on the sts results + # I suspect this is because the wv.vectors are based on the averages of + # wv.vectors_vocab + wv.vectors_ngrams, which will all point into very + # similar directions. + max_ngrams = model.batch_ngrams + w_vectors = model.wv.vectors_vocab + ngram_vectors = model.wv.vectors_ngrams + min_n = model.wv.min_n + max_n = model.wv.max_n + bucket = model.wv.bucket + oov_weight = np_amax(w_weights) + + def get_ft_vector(word:str) -> ndarray: + if word in vocab: + vocab_index = vocab[word].index + return w_vectors[vocab_index] * w_weights[vocab_index] + else: + ngram_hashes = ft_ngram_hashes( + word, min_n, max_n, bucket, True + )[:max_ngrams] + if len(ngram_hashes) == 0: + return zeros(size, dtype=REAL) + return ( + oov_weight * + np_sum(ngram_vectors[ngram_hashes], axis=0) + / len(ngram_hashes) + ) + + eff_sentences, eff_words = 0, 0 + + if not is_ft: + for obj in indexed_sentences: + mem.fill(0.0) + sent = obj[0] + sent_adr = obj[1] + + word_indices = [vocab[word].index for word in sent if word in vocab] + eff_sentences += 1 + if not len(word_indices): + continue + eff_words += len(word_indices) + + if not hierarchical: + # Take the maxium value along the axis + s_vectors[sent_adr] = np_amax( + np_mult(w_vectors[word_indices], w_weights[word_indices][:, None]), + axis=0, + ) + else: + # More expensive iteration + for word_index, _ in enumerate(word_indices): + # Compute the local window + window_indices = word_indices[word_index : word_index + window] + # Perform average pooling [0,1,2,3,4] + mem = np_sum( + np_mult( + w_vectors[window_indices], + w_weights[window_indices][:, None], + ), + axis=0, + ) + # Perform hierarchical max pooling + mem *= 1 / len(window_indices) + s_vectors[sent_adr] = np_maximum(s_vectors[sent_adr], mem,) + else: + for obj in indexed_sentences: + mem.fill(0.0) + sent = obj[0] + sent_adr = obj[1] + + if not len(sent): + continue + mem = zeros(size, dtype=REAL) + + eff_sentences += 1 + eff_words += len(sent) # Counts everything in the sentence + + if not hierarchical: + for word in sent: + if word in vocab: + vocab_index = vocab[word].index + s_vectors[sent_adr] = np_maximum( + get_ft_vector(word), + s_vectors[sent_adr], + ) + else: + ngram_hashes = ft_ngram_hashes( + word, min_n, max_n, bucket, True + )[:max_ngrams] + if len(ngram_hashes) == 0: + continue + mem = oov_weight * ( + np_sum(ngram_vectors[ngram_hashes], axis=0) + / len(ngram_hashes) + ) + + s_vectors[sent_adr] = np_maximum( + mem, + s_vectors[sent_adr], + ) + else: + # Expensive iteration + for word_index, word in enumerate(sent): + mem.fill(0.0) + for context in sent[word_index : word_index + window]: + if word == context: + continue + print(context) + + + + return eff_sentences, eff_words + + +# try: +# from fse.models.average_inner import train_average_cy +# from fse.models.average_inner import ( +# FAST_VERSION, +# MAX_WORDS_IN_BATCH, +# MAX_NGRAMS_IN_BATCH, +# ) + +# train_average = train_average_cy +# except ImportError + +FAST_VERSION = -1 +MAX_WORDS_IN_BATCH = 10000 +MAX_NGRAMS_IN_BATCH = 40 +train_pooling = train_pooling_np + + +class MaxPooling(BaseSentence2VecModel): + """ Train, use and evaluate max pooling sentence vectors. + + The model can be stored/loaded via its :meth:`~fse.models.pooling.MaxPooling.save` and + :meth:`~fse.models.pooling.MaxPooling.load` methods. + + Some important attributes are the following: + + Attributes + ---------- + wv : :class:`~gensim.models.keyedvectors.BaseKeyedVectors` + This object essentially contains the mapping between words and embeddings. After training, it can be used + directly to query those embeddings in various ways. See the module level docstring for examples. + + sv : :class:`~fse.models.sentencevectors.SentenceVectors` + This object contains the sentence vectors inferred from the training data. There will be one such vector + for each unique docusentence supplied during training. They may be individually accessed using the index. + + prep : :class:`~fse.models.base_s2v.BaseSentence2VecPreparer` + The prep object is used to transform and initialize the sv.vectors. Aditionally, it can be used + to move the vectors to disk for training with memmap. + + """ + + def __init__( + self, + model: BaseKeyedVectors, + hierarchical: bool = False, + window_size: int = 5, + sv_mapfile_path: str = None, + wv_mapfile_path: str = None, + workers: int = 1, + lang_freq: str = None, + **kwargs + ): + """ Max pooling sentence embeddings model. Performs a simple maximum pooling operation over all + words in a sentences without further transformations. + + The implementation is based on Shen et al. (2018): Baseline Needs More Love: On Simple Word-Embedding-Based Models and Associated Pooling Mechanisms. + For more information, see . + + Parameters + ---------- + model : :class:`~gensim.models.keyedvectors.BaseKeyedVectors` or :class:`~gensim.models.base_any2vec.BaseWordEmbeddingsModel` + This object essentially contains the mapping between words and embeddings. To compute the sentence embeddings + the wv.vocab and wv.vector elements are required. + hierarchical : bool + If true, then perform hierarchical pooling operation + window_size : int + Set the size of the window used for hierarchical pooling operation + sv_mapfile_path : str, optional + Optional path to store the sentence-vectors in for very large datasets. Used for memmap. + wv_mapfile_path : str, optional + Optional path to store the word-vectors in for very large datasets. Used for memmap. + Use sv_mapfile_path and wv_mapfile_path to train disk-to-disk without needing much ram. + workers : int, optional + Number of working threads, used for multithreading. For most tasks (few words in a sentence) + a value of 1 should be more than enough. + lang_freq : str, optional + Some pre-trained embeddings, i.e. "GoogleNews-vectors-negative300.bin", do not contain information about + the frequency of a word. As the frequency is required for estimating the word weights, we induce + frequencies into the wv.vocab.count based on :class:`~wordfreq` + If no frequency information is available, you can choose the language to estimate the frequency. + See https://github.com/LuminosoInsight/wordfreq + + """ + self.hierarchical = bool(hierarchical) + self.window_size = int(window_size) + + super(MaxPooling, self).__init__( + model=model, + sv_mapfile_path=sv_mapfile_path, + wv_mapfile_path=wv_mapfile_path, + workers=workers, + lang_freq=lang_freq, + batch_words=MAX_WORDS_IN_BATCH, + batch_ngrams=MAX_NGRAMS_IN_BATCH, + fast_version=FAST_VERSION, + ) + + def _do_train_job( + self, data_iterable: List[tuple], target: ndarray, memory: ndarray + ) -> [int, int]: + """ Internal routine which is called on training and performs averaging for all entries in the iterable """ + eff_sentences, eff_words = train_pooling( + model=self, indexed_sentences=data_iterable, target=target, memory=memory, + ) + return eff_sentences, eff_words + + def _check_parameter_sanity(self, **kwargs): + """ Check the sanity of all child paramters """ + if not all(self.word_weights == 1.0): + raise ValueError("All word weights must equal one for pool") + if self.window_size < 1: + raise ValueError("Window size must be greater than 1") + + def _pre_train_calls(self, **kwargs): + """Function calls to perform before training """ + pass + + def _post_train_calls(self, **kwargs): + """ Function calls to perform after training, such as computing eigenvectors """ + pass + + def _post_inference_calls(self, **kwargs): + """ Function calls to perform after training & inference + Examples include the removal of components + """ + pass + + def _check_dtype_santiy(self, **kwargs): + """ Check the dtypes of all child attributes""" + pass diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py new file mode 100644 index 0000000..d6c8ab6 --- /dev/null +++ b/fse/test/test_pooling.py @@ -0,0 +1,326 @@ +#!/usr/bin/env python +# -*- coding: utf-8 -*- + +# Author: Oliver Borchers +# Copyright (C) 2019 Oliver Borchers + +""" +Automated tests for checking the average model. +""" + +import logging +import unittest + +from pathlib import Path + +import numpy as np + +from fse.models.pooling import MaxPooling, train_pooling_np +from fse.models.base_s2v import EPS + +from gensim.models import Word2Vec, FastText + +logger = logging.getLogger(__name__) + +CORPUS = Path("fse/test/test_data/test_sentences.txt") +DIM = 5 +W2V = Word2Vec(min_count=1, size=DIM) +with open(CORPUS, "r") as f: + SENTENCES = [l.split() for i, l in enumerate(f)] +W2V.build_vocab(SENTENCES) +W2V.wv.vectors[:,] = np.arange(len(W2V.wv.vectors), dtype=np.float32)[:, None] + + +class TestAverageFunctions(unittest.TestCase): + def setUp(self): + self.sentences = [ + ["They", "admit"], + ["So", "Apple", "bought", "buds"], + ["go", "12345"], + ["pull", "12345678910111213"], + "this is a longer test sentence test longer sentences".split() + ] + self.sentences = [(s, i) for i, s in enumerate(self.sentences)] + self.model = MaxPooling(W2V) + self.model.prep.prepare_vectors( + sv=self.model.sv, total_sentences=len(self.sentences), update=False + ) + self.model._pre_train_calls() + + # def test_cython(self): + # from fse.models.average_inner import ( + # FAST_VERSION, + # MAX_WORDS_IN_BATCH, + # MAX_NGRAMS_IN_BATCH, + # ) + + # self.assertTrue(FAST_VERSION) + # self.assertEqual(10000, MAX_WORDS_IN_BATCH) + # self.assertEqual(40, MAX_NGRAMS_IN_BATCH) + + # def test_average_train_cy_w2v(self): + # self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + # mem = self.model._get_thread_working_mem() + + # from fse.models.average_inner import train_average_cy + + # output = train_average_cy( + # self.model, self.sentences, self.model.sv.vectors, mem + # ) + # self.assertEqual((4, 7), output) + # self.assertTrue((183 == self.model.sv[0]).all()) + # self.assertTrue((164.5 == self.model.sv[1]).all()) + # self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) + + # def test_average_train_cy_ft(self): + # ft = FastText(min_count=1, size=DIM) + # ft.build_vocab(SENTENCES) + # m = Average(ft) + # m.prep.prepare_vectors( + # sv=m.sv, total_sentences=len(self.sentences), update=False + # ) + # m._pre_train_calls() + # m.wv.vectors = m.wv.vectors_vocab = np.ones_like(m.wv.vectors, dtype=np.float32) + # m.wv.vectors_ngrams = np.full_like(m.wv.vectors_ngrams, 2, dtype=np.float32) + # mem = m._get_thread_working_mem() + + # from fse.models.average_inner import train_average_cy + + # output = train_average_cy(m, self.sentences, m.sv.vectors, mem) + # self.assertEqual((4, 10), output) + # self.assertTrue((1.0 + EPS == m.sv[0]).all()) + # self.assertTrue(np.allclose(1.5, m.sv[2])) + # self.assertTrue(np.allclose(2, m.sv[3])) + + # def test_cy_equal_np_w2v(self): + # m1 = Average(W2V) + # m1.prep.prepare_vectors( + # sv=m1.sv, total_sentences=len(self.sentences), update=False + # ) + # m1._pre_train_calls() + # mem1 = m1._get_thread_working_mem() + # o1 = train_average_np(m1, self.sentences, m1.sv.vectors, mem1) + + # m2 = Average(W2V) + # m2.prep.prepare_vectors( + # sv=m2.sv, total_sentences=len(self.sentences), update=False + # ) + # m2._pre_train_calls() + # mem2 = m2._get_thread_working_mem() + + # from fse.models.average_inner import train_average_cy + + # o2 = train_average_cy(m2, self.sentences, m2.sv.vectors, mem2) + + # self.assertEqual(o1, o2) + # self.assertTrue((m1.sv.vectors == m2.sv.vectors).all()) + + # def test_cy_equal_np_w2v_random(self): + # w2v = Word2Vec(min_count=1, size=DIM) + # # Random initialization + # w2v.build_vocab(SENTENCES) + + # m1 = Average(w2v) + # m1.prep.prepare_vectors( + # sv=m1.sv, total_sentences=len(self.sentences), update=False + # ) + # m1._pre_train_calls() + # mem1 = m1._get_thread_working_mem() + # o1 = train_average_np(m1, self.sentences, m1.sv.vectors, mem1) + + # m2 = Average(w2v) + # m2.prep.prepare_vectors( + # sv=m2.sv, total_sentences=len(self.sentences), update=False + # ) + # m2._pre_train_calls() + # mem2 = m2._get_thread_working_mem() + + # from fse.models.average_inner import train_average_cy + + # o2 = train_average_cy(m2, self.sentences, m2.sv.vectors, mem2) + + # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + # def test_cy_equal_np_ft_random(self): + # ft = FastText(size=20, min_count=1) + # ft.build_vocab(SENTENCES) + + # m1 = Average(ft) + # m1.prep.prepare_vectors( + # sv=m1.sv, total_sentences=len(self.sentences), update=False + # ) + # m1._pre_train_calls() + + # from fse.models.average_inner import MAX_NGRAMS_IN_BATCH + + # m1.batch_ngrams = MAX_NGRAMS_IN_BATCH + # mem1 = m1._get_thread_working_mem() + # o1 = train_average_np(m1, self.sentences[:2], m1.sv.vectors, mem1) + + # m2 = Average(ft) + # m2.prep.prepare_vectors( + # sv=m2.sv, total_sentences=len(self.sentences), update=False + # ) + # m2._pre_train_calls() + # mem2 = m2._get_thread_working_mem() + + # from fse.models.average_inner import train_average_cy + + # o2 = train_average_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) + + # self.assertEqual(o1, o2) + # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + # def test_train_single_from_disk(self): + # p = Path("fse/test/test_data/test_vecs") + # p_res = Path("fse/test/test_data/test_vecs.vectors") + # p_target = Path("fse/test/test_data/test_vecs_wv.vectors") + + # se1 = Average(W2V) + # se2 = Average( + # W2V, sv_mapfile_path=str(p.absolute()), wv_mapfile_path=str(p.absolute()) + # ) + # se1.train([(s, i) for i, s in enumerate(SENTENCES)]) + # se2.train([(s, i) for i, s in enumerate(SENTENCES)]) + + # self.assertTrue(p_target.exists()) + # self.assertTrue((se1.wv.vectors == se2.wv.vectors).all()) + # self.assertFalse(se2.wv.vectors.flags.writeable) + + # self.assertTrue((se1.sv.vectors == se2.sv.vectors).all()) + # p_res.unlink() + # p_target.unlink() + + # def test_train_multi_from_disk(self): + # p = Path("fse/test/test_data/test_vecs") + # p_res = Path("fse/test/test_data/test_vecs.vectors") + # p_target = Path("fse/test/test_data/test_vecs_wv.vectors") + + # se1 = Average(W2V, workers=2) + # se2 = Average( + # W2V, + # workers=2, + # sv_mapfile_path=str(p.absolute()), + # wv_mapfile_path=str(p.absolute()), + # ) + # se1.train([(s, i) for i, s in enumerate(SENTENCES)]) + # se2.train([(s, i) for i, s in enumerate(SENTENCES)]) + + # self.assertTrue(p_target.exists()) + # self.assertTrue((se1.wv.vectors == se2.wv.vectors).all()) + # self.assertFalse(se2.wv.vectors.flags.writeable) + + # self.assertTrue((se1.sv.vectors == se2.sv.vectors).all()) + # p_res.unlink() + # p_target.unlink() + + def test_check_parameter_sanity(self): + se = MaxPooling(W2V) + se.word_weights = np.full(20, 2.0, dtype=np.float32) + with self.assertRaises(ValueError): + se._check_parameter_sanity() + + se = MaxPooling(W2V, window_size=0) + with self.assertRaises(ValueError): + se._check_parameter_sanity() + + def test_train(self): + self.assertEqual( + (100, 1450), self.model.train([(s, i) for i, s in enumerate(SENTENCES)]) + ) + + def test_do_train_job(self): + self.model.prep.prepare_vectors( + sv=self.model.sv, total_sentences=len(SENTENCES), update=True + ) + mem = self.model._get_thread_working_mem() + self.assertEqual( + (100, 1450), + self.model._do_train_job( + [(s, i) for i, s in enumerate(SENTENCES)], + target=self.model.sv.vectors, + memory=mem, + ), + ) + self.assertEqual((105, DIM), self.model.sv.vectors.shape) + + def test_pooling_train_np_w2v(self): + self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + mem = self.model._get_thread_working_mem() + + output = train_pooling_np( + self.model, self.sentences, self.model.sv.vectors, mem + ) + + self.assertEqual((5, 14), output) + self.assertTrue((241 == self.model.sv[0]).all()) + self.assertTrue((306 == self.model.sv[1]).all()) + self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) + + def test_hier_pooling_train_np_w2v(self): + self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + mem = self.model._get_thread_working_mem() + + self.model.hierarchical = True + + output = train_pooling_np( + self.model, self.sentences, self.model.sv.vectors, mem + ) + self.model.hierarchical = False + + self.assertEqual((5, 14), output) + self.assertTrue((183 == self.model.sv[0]).all()) + self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) + + def test_pooling_train_np_ft(self): + ft = FastText(min_count=1, size=DIM) + ft.build_vocab(SENTENCES) + m = MaxPooling(ft) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + + m.wv.vectors[:,] = np.arange(len(m.wv.vectors), dtype=np.float32)[:, None] + m.wv.vectors_vocab = m.wv.vectors + + m.wv.vectors_ngrams[:,] = np.arange(len(m.wv.vectors_ngrams), dtype=np.float32)[:, None] + mem = m._get_thread_working_mem() + + output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + self.assertTrue((241 == m.sv[0]).all()) + self.assertTrue((737413.9 == m.sv[2]).all()) + self.assertTrue((1080970.2 == m.sv[3]).all()) + + def test_hier_pooling_train_np_ft(self): + ft = FastText(min_count=1, size=DIM) + ft.build_vocab(SENTENCES) + m = MaxPooling(ft) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + + m.wv.vectors[:,] = np.arange(len(m.wv.vectors), dtype=np.float32)[:, None] + m.wv.vectors_vocab = m.wv.vectors + + m.wv.vectors_ngrams[:,] = np.arange(len(m.wv.vectors_ngrams), dtype=np.float32)[:, None] + mem = m._get_thread_working_mem() + + m.hierarchical = True + + output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + # self.assertTrue((241 == m.sv[0]).all()) + # self.assertTrue((737413.9 == m.sv[2]).all()) + # self.assertTrue((1080970.2 == m.sv[3]).all()) + + +if __name__ == "__main__": + logging.basicConfig( + format="%(asctime)s : %(levelname)s : %(message)s", level=logging.DEBUG + ) + unittest.main() From fdd9c9b14026739b15e2795a0c19c4192acc0e34 Mon Sep 17 00:00:00 2001 From: OliverB Date: Sun, 16 Feb 2020 10:32:04 +0100 Subject: [PATCH 07/63] Added todo --- fse/models/sentencevectors.py | 25 +++++++++++++++++++++++++ 1 file changed, 25 insertions(+) diff --git a/fse/models/sentencevectors.py b/fse/models/sentencevectors.py index aa5fd95..8d2ddb0 100644 --- a/fse/models/sentencevectors.py +++ b/fse/models/sentencevectors.py @@ -50,6 +50,31 @@ def __init__(self, vector_size: int, mapfile_path: str = None): self.mapfile_path = Path(mapfile_path) if mapfile_path is not None else None self.mapfile_shape = None + """ + Thoughts on additinal features: + ANN: + [ ] Only construct index when when calling most_similar method + [ ] Logging of index speed + [ ] Save and load of index + [ ] Assert that index and vectors are of equal size + [ ] Paramters must be tunable afterwards + [ ] Method to reconstruct index + [ ] How does the index saving comply with SaveLoad? + [ ] Write unittests? + Brute: + [ ] Keep access to default method + [ ] Make ANN Search the default?! --> Results? + [ ] Throw warning for large datasets for vector norm init + [ ] Maybe throw warning if exceeds RAM size of the embedding + normalization + Other: + [ ] L2 Distance + [ ] L1 Distance + [ ] Correlation + [ ] Lookup-Functionality (via defaultdict) + [ ] Get vector: Not really memory friendly + [ ] Show which words are in vocabulary + """ + def __getitem__(self, entities: int) -> ndarray: """Get vector representation of `entities`. From d59241067c291b53712ab2ad21765270d42ad388 Mon Sep 17 00:00:00 2001 From: OliverB Date: Sun, 16 Feb 2020 10:32:15 +0100 Subject: [PATCH 08/63] Fixed n-to-m mapping in np --- fse/models/average.py | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/fse/models/average.py b/fse/models/average.py index d4874e6..83a447d 100644 --- a/fse/models/average.py +++ b/fse/models/average.py @@ -132,7 +132,7 @@ def train_average_np( axis=0, ) mem *= 1 / len(word_indices) - s_vectors[sent_adr] = mem.astype(REAL) + s_vectors[sent_adr] += mem.astype(REAL) else: for obj in indexed_sentences: mem.fill(0.0) @@ -160,7 +160,7 @@ def train_average_np( np_sum(ngram_vectors[ngram_hashes], axis=0) / len(ngram_hashes) ) # Implicit addition of zero if oov does not contain any ngrams - s_vectors[sent_adr] = mem / len(sent) + s_vectors[sent_adr] += mem / len(sent) return eff_sentences, eff_words From a533fa39e30aec397df1b515c3285c3ad176e537 Mon Sep 17 00:00:00 2001 From: OliverB Date: Sun, 16 Feb 2020 11:10:35 +0100 Subject: [PATCH 09/63] Small fix to avg --- fse/models/average.py | 2 +- fse/test/test_average.py | 6 +++--- 2 files changed, 4 insertions(+), 4 deletions(-) diff --git a/fse/models/average.py b/fse/models/average.py index 83a447d..5522d11 100644 --- a/fse/models/average.py +++ b/fse/models/average.py @@ -131,7 +131,7 @@ def train_average_np( np_mult(w_vectors[word_indices], w_weights[word_indices][:, None]), axis=0, ) - mem *= 1 / len(word_indices) + mem /= len(word_indices) s_vectors[sent_adr] += mem.astype(REAL) else: for obj in indexed_sentences: diff --git a/fse/test/test_average.py b/fse/test/test_average.py index 983ab86..0bfd02f 100644 --- a/fse/test/test_average.py +++ b/fse/test/test_average.py @@ -95,9 +95,9 @@ def test_average_train_np_ft(self): mem = m._get_thread_working_mem() output = train_average_np(m, self.sentences, m.sv.vectors, mem) self.assertEqual((4, 10), output) - self.assertTrue((1.0 == m.sv[0]).all()) - self.assertTrue((1.5 == m.sv[2]).all()) - self.assertTrue((2 == m.sv[3]).all()) + self.assertTrue(np.allclose(1.0, m.sv[0])) + self.assertTrue(np.allclose(1.5, m.sv[2])) + self.assertTrue(np.allclose(2, m.sv[3])) # "go" -> [1,1...] # oov: "12345" -> (14 hashes * 2) / 14 = 2 # (2 + 1) / 2 = 1.5 From 83a0c5a267f194b4a11fbeda345393461372286a Mon Sep 17 00:00:00 2001 From: OliverB Date: Sun, 16 Feb 2020 11:11:30 +0100 Subject: [PATCH 10/63] MaxPool work NP code --- fse/models/pooling.py | 54 ++++++++++++++-------------------------- fse/test/test_pooling.py | 14 ++++++++--- 2 files changed, 28 insertions(+), 40 deletions(-) diff --git a/fse/models/pooling.py b/fse/models/pooling.py index 2f5c009..0d788f3 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -122,6 +122,7 @@ def get_ft_vector(word:str) -> ndarray: vocab_index = vocab[word].index return w_vectors[vocab_index] * w_weights[vocab_index] else: + # Requires additional temporary storage ngram_hashes = ft_ngram_hashes( word, min_n, max_n, bucket, True )[:max_ngrams] @@ -167,7 +168,7 @@ def get_ft_vector(word:str) -> ndarray: axis=0, ) # Perform hierarchical max pooling - mem *= 1 / len(window_indices) + mem /= len(window_indices) s_vectors[sent_adr] = np_maximum(s_vectors[sent_adr], mem,) else: for obj in indexed_sentences: @@ -184,41 +185,30 @@ def get_ft_vector(word:str) -> ndarray: if not hierarchical: for word in sent: - if word in vocab: - vocab_index = vocab[word].index - s_vectors[sent_adr] = np_maximum( - get_ft_vector(word), - s_vectors[sent_adr], - ) - else: - ngram_hashes = ft_ngram_hashes( - word, min_n, max_n, bucket, True - )[:max_ngrams] - if len(ngram_hashes) == 0: - continue - mem = oov_weight * ( - np_sum(ngram_vectors[ngram_hashes], axis=0) - / len(ngram_hashes) - ) - - s_vectors[sent_adr] = np_maximum( - mem, - s_vectors[sent_adr], - ) + s_vectors[sent_adr] = np_maximum( + get_ft_vector(word), + s_vectors[sent_adr], + ) else: - # Expensive iteration for word_index, word in enumerate(sent): mem.fill(0.0) + mem += get_ft_vector(word) + count = 1 + for context in sent[word_index : word_index + window]: if word == context: continue - print(context) - + mem += get_ft_vector(context) + count += 1 + mem /= count + s_vectors[sent_adr] = np_maximum( + mem, + s_vectors[sent_adr], + ) return eff_sentences, eff_words - # try: # from fse.models.average_inner import train_average_cy # from fse.models.average_inner import ( @@ -268,7 +258,6 @@ def __init__( sv_mapfile_path: str = None, wv_mapfile_path: str = None, workers: int = 1, - lang_freq: str = None, **kwargs ): """ Max pooling sentence embeddings model. Performs a simple maximum pooling operation over all @@ -293,14 +282,7 @@ def __init__( Use sv_mapfile_path and wv_mapfile_path to train disk-to-disk without needing much ram. workers : int, optional Number of working threads, used for multithreading. For most tasks (few words in a sentence) - a value of 1 should be more than enough. - lang_freq : str, optional - Some pre-trained embeddings, i.e. "GoogleNews-vectors-negative300.bin", do not contain information about - the frequency of a word. As the frequency is required for estimating the word weights, we induce - frequencies into the wv.vocab.count based on :class:`~wordfreq` - If no frequency information is available, you can choose the language to estimate the frequency. - See https://github.com/LuminosoInsight/wordfreq - + a value of 1 should be more than enough. """ self.hierarchical = bool(hierarchical) self.window_size = int(window_size) @@ -310,7 +292,7 @@ def __init__( sv_mapfile_path=sv_mapfile_path, wv_mapfile_path=wv_mapfile_path, workers=workers, - lang_freq=lang_freq, + lang_freq=None, batch_words=MAX_WORDS_IN_BATCH, batch_ngrams=MAX_NGRAMS_IN_BATCH, fast_version=FAST_VERSION, diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index d6c8ab6..f3bd75f 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -314,10 +314,16 @@ def test_hier_pooling_train_np_ft(self): output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) self.assertEqual((5, 19), output) - # self.assertTrue((241 == m.sv[0]).all()) - # self.assertTrue((737413.9 == m.sv[2]).all()) - # self.assertTrue((1080970.2 == m.sv[3]).all()) - + self.assertTrue((183 == m.sv[0]).all()) + self.assertTrue((737413.9 == m.sv[2]).all()) + self.assertTrue((1080970.2 == m.sv[3]).all()) + """ + Note to future self: + Due to the size of the ngram vectors, + an ngram at the last position of the senence + will always be the highest value. + TODO: This unittest is thus a bit flawed. Maybe fix? + """ if __name__ == "__main__": logging.basicConfig( From 5d6e885faa6e03e0ff870d577b101ecd5c75dfb1 Mon Sep 17 00:00:00 2001 From: OliverB Date: Sun, 16 Feb 2020 11:18:11 +0100 Subject: [PATCH 11/63] Typing changes --- fse/models/average.py | 8 ++++---- fse/models/base_s2v.py | 8 ++++---- fse/models/pooling.py | 23 ++++++++++++++++++----- 3 files changed, 26 insertions(+), 13 deletions(-) diff --git a/fse/models/average.py b/fse/models/average.py index 5522d11..02e2adc 100644 --- a/fse/models/average.py +++ b/fse/models/average.py @@ -57,7 +57,7 @@ def train_average_np( model: BaseSentence2VecModel, indexed_sentences: List[tuple], target: ndarray, - memory: ndarray, + memory: tuple, ) -> [int, int]: """Training on a sequence of sentences and update the target ndarray. @@ -77,8 +77,8 @@ def train_average_np( target : ndarray The target ndarray. We use the index from indexed_sentences to write into the corresponding row of target. - memory : ndarray - Private memory for each working thread + memory : tuple + Private memory array(s) for each working thread Returns ------- @@ -254,7 +254,7 @@ def __init__( ) def _do_train_job( - self, data_iterable: List[tuple], target: ndarray, memory: ndarray + self, data_iterable: List[tuple], target: ndarray, memory: tuple ) -> [int, int]: """ Internal routine which is called on training and performs averaging for all entries in the iterable """ eff_sentences, eff_words = train_average( diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index 1846db8..5dc5b84 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -523,13 +523,13 @@ def _move_ndarray_to_disk( readonly_memvecs = np_memmap(path, dtype=REAL, mode="r", shape=shape) return readonly_memvecs - def _get_thread_working_mem(self) -> [ndarray, ndarray]: + def _get_thread_working_mem(self) -> tuple: """Computes the memory used per worker thread. Returns ------- - np.ndarray - Each worker threads private work memory. + tuple + Each worker threads private work memory array(s). """ mem = zeros_aligned(self.sv.vector_size, dtype=REAL) @@ -537,7 +537,7 @@ def _get_thread_working_mem(self) -> [ndarray, ndarray]: return (mem, oov_mem) def _do_train_job( - self, data_iterable: List[tuple], target: ndarray, memory: ndarray + self, data_iterable: List[tuple], target: ndarray, memory: tuple ) -> [int, int]: """ Function to be called on a batch of sentences. Returns eff sentences/words """ raise NotImplementedError() diff --git a/fse/models/pooling.py b/fse/models/pooling.py index 0d788f3..a4bd46d 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -58,7 +58,7 @@ def train_pooling_np( model: BaseSentence2VecModel, indexed_sentences: List[tuple], target: ndarray, - memory: ndarray, + memory: tuple, ) -> [int, int]: """Training on a sequence of sentences and update the target ndarray. @@ -67,7 +67,7 @@ def train_pooling_np( Warnings -------- This is the non-optimized, pure Python version. If you have a C compiler, - fse will use an optimized code path from :mod:`fse.models.average_inner` instead. + fse will use an optimized code path from :mod:`fse.models.pooling_inner` instead. Parameters ---------- @@ -78,8 +78,8 @@ def train_pooling_np( target : ndarray The target ndarray. We use the index from indexed_sentences to write into the corresponding row of target. - memory : ndarray - Private memory for each working thread + memory : tuple + Private memory array(s) for each working thread Returns ------- @@ -118,6 +118,19 @@ def train_pooling_np( oov_weight = np_amax(w_weights) def get_ft_vector(word:str) -> ndarray: + """ Function to compute the FT vectors if applicable + + Parameters + ---------- + word : str + String representation of token + + Returns + ------- + ndarray + FT vector representation + + """ if word in vocab: vocab_index = vocab[word].index return w_vectors[vocab_index] * w_weights[vocab_index] @@ -299,7 +312,7 @@ def __init__( ) def _do_train_job( - self, data_iterable: List[tuple], target: ndarray, memory: ndarray + self, data_iterable: List[tuple], target: ndarray, memory: tuple ) -> [int, int]: """ Internal routine which is called on training and performs averaging for all entries in the iterable """ eff_sentences, eff_words = train_pooling( From 00b33cb75fec3fc2bf41e47b2b28aa50fc9110f0 Mon Sep 17 00:00:00 2001 From: OliverB Date: Sun, 16 Feb 2020 11:47:40 +0100 Subject: [PATCH 12/63] Added pooling --- fse/models/__init__.py | 3 ++- 1 file changed, 2 insertions(+), 1 deletion(-) diff --git a/fse/models/__init__.py b/fse/models/__init__.py index 3f13790..38f90b4 100644 --- a/fse/models/__init__.py +++ b/fse/models/__init__.py @@ -1,4 +1,5 @@ from .average import Average from .sif import SIF from .usif import uSIF -from .sentencevectors import SentenceVectors +from .pooling import MaxPooling +from .sentencevectors import SentenceVectors \ No newline at end of file From 976df4460ff297d2ab89b9a820a1aea3f0f44aec Mon Sep 17 00:00:00 2001 From: OliverB Date: Sun, 16 Feb 2020 11:47:51 +0100 Subject: [PATCH 13/63] Fixed non-negative bug + tests --- fse/models/pooling.py | 3 ++- fse/test/test_pooling.py | 49 +++++++++++++++++++++++++++------------- 2 files changed, 35 insertions(+), 17 deletions(-) diff --git a/fse/models/pooling.py b/fse/models/pooling.py index a4bd46d..470be00 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -163,10 +163,11 @@ def get_ft_vector(word:str) -> ndarray: if not hierarchical: # Take the maxium value along the axis - s_vectors[sent_adr] = np_amax( + mem = np_amax( np_mult(w_vectors[word_indices], w_weights[word_indices][:, None]), axis=0, ) + s_vectors[sent_adr] = np_maximum(s_vectors[sent_adr], mem,) else: # More expensive iteration for word_index, _ in enumerate(word_indices): diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index f3bd75f..1503a53 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -30,6 +30,17 @@ W2V.build_vocab(SENTENCES) W2V.wv.vectors[:,] = np.arange(len(W2V.wv.vectors), dtype=np.float32)[:, None] +W2V_R = Word2Vec(min_count=1, size=DIM) +W2V_R.build_vocab(SENTENCES) + +FT = FastText(min_count=1, size=DIM) +FT.build_vocab(SENTENCES) +FT.wv.vectors[:,] = np.arange(len(FT.wv.vectors), dtype=np.float32)[:, None] +FT.wv.vectors_vocab = FT.wv.vectors +FT.wv.vectors_ngrams[:,] = np.arange(len(FT.wv.vectors_ngrams), dtype=np.float32)[:, None] + +FT_R = FastText(min_count=1, size=DIM) +FT_R.build_vocab(SENTENCES) class TestAverageFunctions(unittest.TestCase): def setUp(self): @@ -257,6 +268,11 @@ def test_pooling_train_np_w2v(self): self.assertTrue((306 == self.model.sv[1]).all()) self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) + def test_pooling_train_np_w2v_non_negative(self): + mpool = MaxPooling(W2V_R) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) + def test_hier_pooling_train_np_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() @@ -272,19 +288,17 @@ def test_hier_pooling_train_np_w2v(self): self.assertTrue((183 == self.model.sv[0]).all()) self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) + def test_hpooling_train_np_w2v_non_negative(self): + mpool = MaxPooling(W2V_R, hierarchical=True) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) + def test_pooling_train_np_ft(self): - ft = FastText(min_count=1, size=DIM) - ft.build_vocab(SENTENCES) - m = MaxPooling(ft) + m = MaxPooling(FT) m.prep.prepare_vectors( sv=m.sv, total_sentences=len(self.sentences), update=False ) m._pre_train_calls() - - m.wv.vectors[:,] = np.arange(len(m.wv.vectors), dtype=np.float32)[:, None] - m.wv.vectors_vocab = m.wv.vectors - - m.wv.vectors_ngrams[:,] = np.arange(len(m.wv.vectors_ngrams), dtype=np.float32)[:, None] mem = m._get_thread_working_mem() output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) @@ -294,19 +308,17 @@ def test_pooling_train_np_ft(self): self.assertTrue((737413.9 == m.sv[2]).all()) self.assertTrue((1080970.2 == m.sv[3]).all()) + def test_pooling_train_np_ft_non_negative(self): + mpool = MaxPooling(FT_R) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) + def test_hier_pooling_train_np_ft(self): - ft = FastText(min_count=1, size=DIM) - ft.build_vocab(SENTENCES) - m = MaxPooling(ft) + m = MaxPooling(FT) m.prep.prepare_vectors( sv=m.sv, total_sentences=len(self.sentences), update=False ) m._pre_train_calls() - - m.wv.vectors[:,] = np.arange(len(m.wv.vectors), dtype=np.float32)[:, None] - m.wv.vectors_vocab = m.wv.vectors - - m.wv.vectors_ngrams[:,] = np.arange(len(m.wv.vectors_ngrams), dtype=np.float32)[:, None] mem = m._get_thread_working_mem() m.hierarchical = True @@ -325,6 +337,11 @@ def test_hier_pooling_train_np_ft(self): TODO: This unittest is thus a bit flawed. Maybe fix? """ + def test_hier_pooling_train_np_ft_non_negative(self): + mpool = MaxPooling(FT_R, hierarchical=True) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) + if __name__ == "__main__": logging.basicConfig( format="%(asctime)s : %(levelname)s : %(message)s", level=logging.DEBUG From fb1d55b9afdb064f3dcc3a6dcbefb51d09a867ca Mon Sep 17 00:00:00 2001 From: OliverB Date: Sun, 16 Feb 2020 12:16:28 +0100 Subject: [PATCH 14/63] Updated readme --- README.md | 31 +++++++++++++++++++------------ 1 file changed, 19 insertions(+), 12 deletions(-) diff --git a/README.md b/README.md index aff4a04..986f12f 100644 --- a/README.md +++ b/README.md @@ -20,16 +20,14 @@ Find the corresponding blog post(s) here: - [Visualizing 100,000 Amazon Products](https://towardsdatascience.com/vis-amz-83dea6fcb059) - [Sentence Embeddings. Fast, please!](https://towardsdatascience.com/fse-2b1ffa791cf9) -- **Announcment: Please understand, that I am at the end of my PhD and I do not have many free minutes to fix issues or add features.** - -**fse** implements three algorithms for sentence embeddings. You can choose -between *unweighted sentence averages*, *smooth inverse frequency averages*, and *unsupervised smooth inverse frequency averages*. +**fse** implements five algorithms for sentence embeddings. You can choose +between *unweighted sentence averages*, *smooth inverse frequency averages*, *unsupervised smooth inverse frequency averages*, *max pooling*, and *hierarchical max pooling*. Key features of **fse** are: **[X]** Up to 500.000 sentences / second (1) -**[X]** Supports Average, SIF, and uSIF Embeddings +**[X]** Supports Average, SIF, uSIF, MaxPooling, and Hierarchical MaxPooling Embeddings **[X]** Full support for Gensims Word2Vec and all other compatible classes @@ -51,19 +49,14 @@ Key features of **fse** are: **[X]** Extensive documentation of all functions +**[X]** Extensive unittest for Linux/OSX + **[X]** Optimized Input Classes (1) May vary significantly from system to system (i.e. by using swap memory) and processing. I regularly observe 300k-500k sentences/s for preprocessed data on my Macbook (2016). Visit **Tutorial.ipynb** for an example. -Things I will work on next: - -**[ ]** MaxPooling / Hierarchical Pooling Embedding - -**[ ]** Approximate Nearest Neighbor Search for SentenceVectors - - Installation ------------ @@ -107,6 +100,7 @@ The models presented are based on - Deep-averaging embeddings [1] - Smooth inverse frequency embeddings [2] - Unsupervised smooth inverse frequency embeddings [3] +- MaxPooling / Hierarchical MaxPooling [5] Credits to Radim Řehůřek and all contributors for the **awesome** library and code that [Gensim](https://github.com/RaRe-Technologies/gensim) provides. A whole lot of the code found in this lib is based on Gensim. @@ -156,6 +150,17 @@ Model | [STS Benchmark](http://ixa2.si.ehu.es/stswiki/index.php/STSbenchmark#Re Changelog ------------- +0.1.16 from 0.1.15: +- Added MaxPooling / Hierarchical MaxPooling +- Added features to Sentencevectors +- Added further unittests +- Workaround for Numpy memmap issue (https://github.com/numpy/numpy/issues/13172) +- Bugfixes for python 3.8 builds +- Code refactoring to black style +- SVD ram subsampling for SIF / uSIF +- Minor fixes for nan-handling +- Minor fix for sentencevectors class + 0.1.15 from 0.1.11: - Fixed major FT Ngram computation bug - Rewrote the input class. Turns out NamedTuple was pretty slow. @@ -181,6 +186,8 @@ Proceedings of the 3rd Workshop on Representation Learning for NLP. (Toulon, Fra 4. Eneko Agirre, Daniel Cer, Mona Diab, Iñigo Lopez-Gazpio, Lucia Specia. Semeval-2017 Task 1: Semantic Textual Similarity Multilingual and Crosslingual Focused Evaluation. Proceedings of SemEval 2017. +5. Dinghan Shen, Guoyin Wang, Wenlin Wang, Martin Renqiang Min, Qinliang Su, Yizhe Zhang, Chunyuan Li, Ricardo Henao, Lawrence Carin (2018) Baseline Needs More Love: On Simple Word-Embedding-Based Models and Associated Pooling Mechanisms. ACL 2018. + Copyright ------------- From e8a7bd990566cacd8bb4c72963a8e90009fe8e38 Mon Sep 17 00:00:00 2001 From: OliverB Date: Sun, 16 Feb 2020 13:04:43 +0100 Subject: [PATCH 15/63] Corrected comments --- fse/models/pooling.py | 2 +- fse/test/test_pooling.py | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/fse/models/pooling.py b/fse/models/pooling.py index 470be00..93851d1 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -315,7 +315,7 @@ def __init__( def _do_train_job( self, data_iterable: List[tuple], target: ndarray, memory: tuple ) -> [int, int]: - """ Internal routine which is called on training and performs averaging for all entries in the iterable """ + """ Internal routine which is called on training and performs pooling for all entries in the iterable """ eff_sentences, eff_words = train_pooling( model=self, indexed_sentences=data_iterable, target=target, memory=memory, ) diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index 1503a53..1179bbf 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -5,7 +5,7 @@ # Copyright (C) 2019 Oliver Borchers """ -Automated tests for checking the average model. +Automated tests for checking the model. """ import logging From 5c94d5cff5513da785c7bbd16713da6658024c78 Mon Sep 17 00:00:00 2001 From: OliverB Date: Sun, 16 Feb 2020 13:06:55 +0100 Subject: [PATCH 16/63] Black formatting --- fse/models/__init__.py | 2 +- fse/models/pooling.py | 21 +++++++++------------ fse/test/test_inputs.py | 15 +++++++++------ fse/test/test_pooling.py | 18 +++++++++++------- 4 files changed, 30 insertions(+), 26 deletions(-) diff --git a/fse/models/__init__.py b/fse/models/__init__.py index 38f90b4..d3e0f92 100644 --- a/fse/models/__init__.py +++ b/fse/models/__init__.py @@ -2,4 +2,4 @@ from .sif import SIF from .usif import uSIF from .pooling import MaxPooling -from .sentencevectors import SentenceVectors \ No newline at end of file +from .sentencevectors import SentenceVectors diff --git a/fse/models/pooling.py b/fse/models/pooling.py index 93851d1..a14f4e4 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -117,7 +117,7 @@ def train_pooling_np( bucket = model.wv.bucket oov_weight = np_amax(w_weights) - def get_ft_vector(word:str) -> ndarray: + def get_ft_vector(word: str) -> ndarray: """ Function to compute the FT vectors if applicable Parameters @@ -136,14 +136,14 @@ def get_ft_vector(word:str) -> ndarray: return w_vectors[vocab_index] * w_weights[vocab_index] else: # Requires additional temporary storage - ngram_hashes = ft_ngram_hashes( - word, min_n, max_n, bucket, True - )[:max_ngrams] + ngram_hashes = ft_ngram_hashes(word, min_n, max_n, bucket, True)[ + :max_ngrams + ] if len(ngram_hashes) == 0: return zeros(size, dtype=REAL) return ( - oov_weight * - np_sum(ngram_vectors[ngram_hashes], axis=0) + oov_weight + * np_sum(ngram_vectors[ngram_hashes], axis=0) / len(ngram_hashes) ) @@ -200,8 +200,7 @@ def get_ft_vector(word:str) -> ndarray: if not hierarchical: for word in sent: s_vectors[sent_adr] = np_maximum( - get_ft_vector(word), - s_vectors[sent_adr], + get_ft_vector(word), s_vectors[sent_adr], ) else: for word_index, word in enumerate(sent): @@ -216,13 +215,11 @@ def get_ft_vector(word:str) -> ndarray: count += 1 mem /= count - s_vectors[sent_adr] = np_maximum( - mem, - s_vectors[sent_adr], - ) + s_vectors[sent_adr] = np_maximum(mem, s_vectors[sent_adr],) return eff_sentences, eff_words + # try: # from fse.models.average_inner import train_average_cy # from fse.models.average_inner import ( diff --git a/fse/test/test_inputs.py b/fse/test/test_inputs.py index 97bef70..7aec523 100644 --- a/fse/test/test_inputs.py +++ b/fse/test/test_inputs.py @@ -134,7 +134,7 @@ def test_mutable_funcs(self): self.il.insert(0, "the") with self.assertRaises(NotImplementedError): self.il.append("the") - + def test_arg_merging(self): with self.assertRaises(RuntimeError): CIndexedList(self.list_a, self.list_a, custom_index=[1, 1]) @@ -170,7 +170,7 @@ def test_mutable_funcs(self): self.il.insert(0, "the") with self.assertRaises(NotImplementedError): self.il.append("the") - + def test_arg_merging(self): with self.assertRaises(RuntimeError): SplitCIndexedList(self.list_a, self.list_a, custom_index=[1, 1]) @@ -199,14 +199,17 @@ def test_mutable_funcs(self): self.il.insert(0, "the") with self.assertRaises(NotImplementedError): self.il.append("the") - + def test_arg_merging(self): with self.assertRaises(RuntimeError): - CSplitCIndexedList(self.list_a, self.list_a, + CSplitCIndexedList( + self.list_a, + self.list_a, custom_split=self.split_func, - custom_index=[1, 1] + custom_index=[1, 1], ) + class TestIndexedLineDocument(unittest.TestCase): def setUp(self): self.p = "fse/test/test_data/test_sentences.txt" @@ -233,4 +236,4 @@ def test_yield(self): logging.basicConfig( format="%(asctime)s : %(levelname)s : %(message)s", level=logging.DEBUG ) - unittest.main() \ No newline at end of file + unittest.main() diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index 1179bbf..519df98 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -37,11 +37,14 @@ FT.build_vocab(SENTENCES) FT.wv.vectors[:,] = np.arange(len(FT.wv.vectors), dtype=np.float32)[:, None] FT.wv.vectors_vocab = FT.wv.vectors -FT.wv.vectors_ngrams[:,] = np.arange(len(FT.wv.vectors_ngrams), dtype=np.float32)[:, None] +FT.wv.vectors_ngrams[:,] = np.arange(len(FT.wv.vectors_ngrams), dtype=np.float32)[ + :, None +] FT_R = FastText(min_count=1, size=DIM) FT_R.build_vocab(SENTENCES) + class TestAverageFunctions(unittest.TestCase): def setUp(self): self.sentences = [ @@ -49,7 +52,7 @@ def setUp(self): ["So", "Apple", "bought", "buds"], ["go", "12345"], ["pull", "12345678910111213"], - "this is a longer test sentence test longer sentences".split() + "this is a longer test sentence test longer sentences".split(), ] self.sentences = [(s, i) for i, s in enumerate(self.sentences)] self.model = MaxPooling(W2V) @@ -230,16 +233,16 @@ def test_check_parameter_sanity(self): se.word_weights = np.full(20, 2.0, dtype=np.float32) with self.assertRaises(ValueError): se._check_parameter_sanity() - + se = MaxPooling(W2V, window_size=0) with self.assertRaises(ValueError): se._check_parameter_sanity() - + def test_train(self): self.assertEqual( (100, 1450), self.model.train([(s, i) for i, s in enumerate(SENTENCES)]) ) - + def test_do_train_job(self): self.model.prep.prepare_vectors( sv=self.model.sv, total_sentences=len(SENTENCES), update=True @@ -272,13 +275,13 @@ def test_pooling_train_np_w2v_non_negative(self): mpool = MaxPooling(W2V_R) mpool.train(self.sentences) self.assertTrue((mpool.sv.vectors >= 0).all()) - + def test_hier_pooling_train_np_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() self.model.hierarchical = True - + output = train_pooling_np( self.model, self.sentences, self.model.sv.vectors, mem ) @@ -342,6 +345,7 @@ def test_hier_pooling_train_np_ft_non_negative(self): mpool.train(self.sentences) self.assertTrue((mpool.sv.vectors >= 0).all()) + if __name__ == "__main__": logging.basicConfig( format="%(asctime)s : %(levelname)s : %(message)s", level=logging.DEBUG From 727efb2621e6f03cf0d6c85011463e1d5e227a5f Mon Sep 17 00:00:00 2001 From: OliverB Date: Tue, 18 Feb 2020 12:08:07 +0100 Subject: [PATCH 17/63] Updated readme --- README.md | 40 +++++++++++++++++++++------------------- 1 file changed, 21 insertions(+), 19 deletions(-) diff --git a/README.md b/README.md index 986f12f..7754ea1 100644 --- a/README.md +++ b/README.md @@ -81,6 +81,25 @@ If building the Cython extension fails (you will be notified), try: Usage ------------- +In order to use **fse** you must first estimate a Gensim model which contains a +gensim.models.keyedvectors.BaseKeyedVectors class, for example +*Word2Vec* or *Fasttext*. Then you can proceed to compute sentence embeddings +for a corpus. + + from gensim.models import FastText + sentences = [["cat", "say", "meow"], ["dog", "say", "woof"]] + ft = FastText(sentences, min_count=1, size=10) + + from fse.models import Average + from fse import IndexedList + model = Average(ft) + model.train(IndexedList(sentences)) + + model.sv.similarity(0,1) + +fse offers multi-thread support out of the box. However, for most +applications a *single thread will most likely be sufficient*. + Within the folder nootebooks you can find the following guides: **Tutorial.ipynb** offers a detailed walk-through of some of the most important functions fse has to offer. @@ -105,30 +124,13 @@ The models presented are based on Credits to Radim Řehůřek and all contributors for the **awesome** library and code that [Gensim](https://github.com/RaRe-Technologies/gensim) provides. A whole lot of the code found in this lib is based on Gensim. -In order to use **fse** you must first estimate a Gensim model which contains a -gensim.models.keyedvectors.BaseKeyedVectors class, for example -*Word2Vec* or *Fasttext*. Then you can proceed to compute sentence embeddings -for a corpus. - - from gensim.models import FastText - sentences = [["cat", "say", "meow"], ["dog", "say", "woof"]] - ft = FastText(sentences, min_count=1, size=10) - - from fse.models import Average - from fse import IndexedList - model = Average(ft) - model.train(IndexedList(sentences)) - - model.sv.similarity(0,1) - -fse offers multi-thread support out of the box. However, for most -applications a *single thread will most likely be sufficient*. - To install **fse** on Colab, check out: https://colab.research.google.com/drive/1qq9GBgEosG7YSRn7r6e02T9snJb04OEi Results ------------ +Note: Though some models perform very good on the sentence-similarty-task (STS), this does not imply good performance in other donwstream tasks! + Model | [STS Benchmark](http://ixa2.si.ehu.es/stswiki/index.php/STSbenchmark#Results) :---: | :---: `CBOW-Paranmt` | **79.85** From 4f936d8f6f2986af8b7faaf5c3c34a0235fd9d40 Mon Sep 17 00:00:00 2001 From: OliverB Date: Tue, 18 Feb 2020 12:10:55 +0100 Subject: [PATCH 18/63] Updated readme --- README.md | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) diff --git a/README.md b/README.md index 7754ea1..8fcde07 100644 --- a/README.md +++ b/README.md @@ -78,6 +78,8 @@ If building the Cython extension fails (you will be notified), try: pip install -U git+https://github.com/oborchers/Fast_Sentence_Embeddings +To install **fse** on Colab, check out: https://colab.research.google.com/drive/1qq9GBgEosG7YSRn7r6e02T9snJb04OEi + Usage ------------- @@ -124,8 +126,6 @@ The models presented are based on Credits to Radim Řehůřek and all contributors for the **awesome** library and code that [Gensim](https://github.com/RaRe-Technologies/gensim) provides. A whole lot of the code found in this lib is based on Gensim. -To install **fse** on Colab, check out: https://colab.research.google.com/drive/1qq9GBgEosG7YSRn7r6e02T9snJb04OEi - Results ------------ @@ -159,7 +159,7 @@ Changelog - Workaround for Numpy memmap issue (https://github.com/numpy/numpy/issues/13172) - Bugfixes for python 3.8 builds - Code refactoring to black style -- SVD ram subsampling for SIF / uSIF +- SVD ram subsampling for SIF / uSIF (customizable, standard is 1 GB of RAM) - Minor fixes for nan-handling - Minor fix for sentencevectors class From cc31045ed60fdac16f2fd4270fe4b993aa56463a Mon Sep 17 00:00:00 2001 From: OliverB Date: Tue, 18 Feb 2020 12:14:57 +0100 Subject: [PATCH 19/63] Added todos --- fse/models/base_s2v.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index 5dc5b84..3dc4b42 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -131,13 +131,14 @@ def __init__( [ ] windows support [ ] documentation [ ] more benchmarks + [ ] more speed benchmarks -> worker support still somewhat weird [ ] remove wv_mapfile_path? [ ] modifiable sv_mapfile_path? [ ] models: [ ] check feasibility first - [ ] max-pooling -> easy - [ ] hierarchical pooling -> easy + [X] max-pooling -> easy + [X] hierarchical pooling -> easy [ ] discrete cosine transform -> somewhat easy, questionable [ ] valve -> unclear, not cited enough [ ] power-means embedding -> very large dimensionalty From 1b144aca213b483682e23dbabd26a82d54bd6d19 Mon Sep 17 00:00:00 2001 From: OliverB Date: Wed, 26 Feb 2020 17:57:24 +0100 Subject: [PATCH 20/63] First MaxPool Implementation --- fse/models/average_inner.c | 10722 +++++++++++++++++++++++++++++++++ fse/models/average_inner.pxd | 12 +- fse/models/average_inner.pyx | 7 +- fse/models/base_s2v.py | 9 +- fse/models/pooling.py | 30 +- fse/models/pooling_inner.c | 8350 +++++++++++++++++++++++++ fse/models/pooling_inner.pyx | 211 + fse/test/test_base_s2v.py | 7 + fse/test/test_pooling.py | 154 +- 9 files changed, 19406 insertions(+), 96 deletions(-) create mode 100644 fse/models/average_inner.c create mode 100644 fse/models/pooling_inner.c create mode 100644 fse/models/pooling_inner.pyx diff --git a/fse/models/average_inner.c b/fse/models/average_inner.c new file mode 100644 index 0000000..91211b6 --- /dev/null +++ b/fse/models/average_inner.c @@ -0,0 +1,10722 @@ +/* Generated by Cython 0.29.14 */ + +/* BEGIN: Cython Metadata +{ + "distutils": { + "depends": [ + "voidptr.h" + ], + "extra_compile_args": [ + "-O2", + "-march=native" + ], + "name": "average_inner", + "sources": [ + "average_inner.pyx" + ] + }, + "module_name": "average_inner" +} +END: Cython Metadata */ + +#define PY_SSIZE_T_CLEAN +#include "Python.h" +#ifndef Py_PYTHON_H + #error Python headers needed to compile C extensions, please install development version of Python. +#elif PY_VERSION_HEX < 0x02060000 || (0x03000000 <= PY_VERSION_HEX && PY_VERSION_HEX < 0x03030000) + #error Cython requires Python 2.6+ or Python 3.3+. +#else +#define CYTHON_ABI "0_29_14" +#define CYTHON_HEX_VERSION 0x001D0EF0 +#define CYTHON_FUTURE_DIVISION 0 +#include +#ifndef offsetof + #define offsetof(type, member) ( (size_t) & ((type*)0) -> member ) +#endif +#if !defined(WIN32) && !defined(MS_WINDOWS) + #ifndef __stdcall + #define __stdcall + #endif + #ifndef __cdecl + #define __cdecl + #endif + #ifndef __fastcall + #define __fastcall + #endif +#endif +#ifndef DL_IMPORT + #define DL_IMPORT(t) t +#endif +#ifndef DL_EXPORT + #define DL_EXPORT(t) t +#endif +#define __PYX_COMMA , +#ifndef HAVE_LONG_LONG + #if PY_VERSION_HEX >= 0x02070000 + #define HAVE_LONG_LONG + #endif +#endif +#ifndef PY_LONG_LONG + #define PY_LONG_LONG LONG_LONG +#endif +#ifndef Py_HUGE_VAL + #define Py_HUGE_VAL HUGE_VAL +#endif +#ifdef PYPY_VERSION + #define CYTHON_COMPILING_IN_PYPY 1 + #define CYTHON_COMPILING_IN_PYSTON 0 + #define CYTHON_COMPILING_IN_CPYTHON 0 + #undef CYTHON_USE_TYPE_SLOTS + #define CYTHON_USE_TYPE_SLOTS 0 + #undef CYTHON_USE_PYTYPE_LOOKUP + #define CYTHON_USE_PYTYPE_LOOKUP 0 + #if PY_VERSION_HEX < 0x03050000 + #undef CYTHON_USE_ASYNC_SLOTS + #define CYTHON_USE_ASYNC_SLOTS 0 + #elif !defined(CYTHON_USE_ASYNC_SLOTS) + #define CYTHON_USE_ASYNC_SLOTS 1 + #endif + #undef CYTHON_USE_PYLIST_INTERNALS + #define CYTHON_USE_PYLIST_INTERNALS 0 + #undef CYTHON_USE_UNICODE_INTERNALS + #define CYTHON_USE_UNICODE_INTERNALS 0 + #undef CYTHON_USE_UNICODE_WRITER + #define CYTHON_USE_UNICODE_WRITER 0 + #undef CYTHON_USE_PYLONG_INTERNALS + #define CYTHON_USE_PYLONG_INTERNALS 0 + #undef CYTHON_AVOID_BORROWED_REFS + #define CYTHON_AVOID_BORROWED_REFS 1 + #undef CYTHON_ASSUME_SAFE_MACROS + #define CYTHON_ASSUME_SAFE_MACROS 0 + #undef CYTHON_UNPACK_METHODS + #define CYTHON_UNPACK_METHODS 0 + #undef CYTHON_FAST_THREAD_STATE + #define CYTHON_FAST_THREAD_STATE 0 + #undef CYTHON_FAST_PYCALL + #define CYTHON_FAST_PYCALL 0 + #undef CYTHON_PEP489_MULTI_PHASE_INIT + #define CYTHON_PEP489_MULTI_PHASE_INIT 0 + #undef CYTHON_USE_TP_FINALIZE + #define CYTHON_USE_TP_FINALIZE 0 + #undef CYTHON_USE_DICT_VERSIONS + #define CYTHON_USE_DICT_VERSIONS 0 + #undef CYTHON_USE_EXC_INFO_STACK + #define CYTHON_USE_EXC_INFO_STACK 0 +#elif defined(PYSTON_VERSION) + #define CYTHON_COMPILING_IN_PYPY 0 + #define CYTHON_COMPILING_IN_PYSTON 1 + #define CYTHON_COMPILING_IN_CPYTHON 0 + #ifndef CYTHON_USE_TYPE_SLOTS + #define CYTHON_USE_TYPE_SLOTS 1 + #endif + #undef CYTHON_USE_PYTYPE_LOOKUP + #define CYTHON_USE_PYTYPE_LOOKUP 0 + #undef CYTHON_USE_ASYNC_SLOTS + #define CYTHON_USE_ASYNC_SLOTS 0 + #undef CYTHON_USE_PYLIST_INTERNALS + #define CYTHON_USE_PYLIST_INTERNALS 0 + #ifndef CYTHON_USE_UNICODE_INTERNALS + #define CYTHON_USE_UNICODE_INTERNALS 1 + #endif + #undef CYTHON_USE_UNICODE_WRITER + #define CYTHON_USE_UNICODE_WRITER 0 + #undef CYTHON_USE_PYLONG_INTERNALS + #define CYTHON_USE_PYLONG_INTERNALS 0 + #ifndef CYTHON_AVOID_BORROWED_REFS + #define CYTHON_AVOID_BORROWED_REFS 0 + #endif + #ifndef CYTHON_ASSUME_SAFE_MACROS + #define CYTHON_ASSUME_SAFE_MACROS 1 + #endif + #ifndef CYTHON_UNPACK_METHODS + #define CYTHON_UNPACK_METHODS 1 + #endif + #undef CYTHON_FAST_THREAD_STATE + #define CYTHON_FAST_THREAD_STATE 0 + #undef CYTHON_FAST_PYCALL + #define CYTHON_FAST_PYCALL 0 + #undef CYTHON_PEP489_MULTI_PHASE_INIT + #define CYTHON_PEP489_MULTI_PHASE_INIT 0 + #undef CYTHON_USE_TP_FINALIZE + #define CYTHON_USE_TP_FINALIZE 0 + #undef CYTHON_USE_DICT_VERSIONS + #define CYTHON_USE_DICT_VERSIONS 0 + #undef CYTHON_USE_EXC_INFO_STACK + #define CYTHON_USE_EXC_INFO_STACK 0 +#else + #define CYTHON_COMPILING_IN_PYPY 0 + #define CYTHON_COMPILING_IN_PYSTON 0 + #define CYTHON_COMPILING_IN_CPYTHON 1 + #ifndef CYTHON_USE_TYPE_SLOTS + #define CYTHON_USE_TYPE_SLOTS 1 + #endif + #if PY_VERSION_HEX < 0x02070000 + #undef CYTHON_USE_PYTYPE_LOOKUP + #define CYTHON_USE_PYTYPE_LOOKUP 0 + #elif !defined(CYTHON_USE_PYTYPE_LOOKUP) + #define CYTHON_USE_PYTYPE_LOOKUP 1 + #endif + #if PY_MAJOR_VERSION < 3 + #undef CYTHON_USE_ASYNC_SLOTS + #define CYTHON_USE_ASYNC_SLOTS 0 + #elif !defined(CYTHON_USE_ASYNC_SLOTS) + #define CYTHON_USE_ASYNC_SLOTS 1 + #endif + #if PY_VERSION_HEX < 0x02070000 + #undef CYTHON_USE_PYLONG_INTERNALS + #define CYTHON_USE_PYLONG_INTERNALS 0 + #elif !defined(CYTHON_USE_PYLONG_INTERNALS) + #define CYTHON_USE_PYLONG_INTERNALS 1 + #endif + #ifndef CYTHON_USE_PYLIST_INTERNALS + #define CYTHON_USE_PYLIST_INTERNALS 1 + #endif + #ifndef CYTHON_USE_UNICODE_INTERNALS + #define CYTHON_USE_UNICODE_INTERNALS 1 + #endif + #if PY_VERSION_HEX < 0x030300F0 + #undef CYTHON_USE_UNICODE_WRITER + #define CYTHON_USE_UNICODE_WRITER 0 + #elif !defined(CYTHON_USE_UNICODE_WRITER) + #define CYTHON_USE_UNICODE_WRITER 1 + #endif + #ifndef CYTHON_AVOID_BORROWED_REFS + #define CYTHON_AVOID_BORROWED_REFS 0 + #endif + #ifndef CYTHON_ASSUME_SAFE_MACROS + #define CYTHON_ASSUME_SAFE_MACROS 1 + #endif + #ifndef CYTHON_UNPACK_METHODS + #define CYTHON_UNPACK_METHODS 1 + #endif + #ifndef CYTHON_FAST_THREAD_STATE + #define CYTHON_FAST_THREAD_STATE 1 + #endif + #ifndef CYTHON_FAST_PYCALL + #define CYTHON_FAST_PYCALL 1 + #endif + #ifndef CYTHON_PEP489_MULTI_PHASE_INIT + #define CYTHON_PEP489_MULTI_PHASE_INIT (PY_VERSION_HEX >= 0x03050000) + #endif + #ifndef CYTHON_USE_TP_FINALIZE + #define CYTHON_USE_TP_FINALIZE (PY_VERSION_HEX >= 0x030400a1) + #endif + #ifndef CYTHON_USE_DICT_VERSIONS + #define CYTHON_USE_DICT_VERSIONS (PY_VERSION_HEX >= 0x030600B1) + #endif + #ifndef CYTHON_USE_EXC_INFO_STACK + #define CYTHON_USE_EXC_INFO_STACK (PY_VERSION_HEX >= 0x030700A3) + #endif +#endif +#if !defined(CYTHON_FAST_PYCCALL) +#define CYTHON_FAST_PYCCALL (CYTHON_FAST_PYCALL && PY_VERSION_HEX >= 0x030600B1) +#endif +#if CYTHON_USE_PYLONG_INTERNALS + #include "longintrepr.h" + #undef SHIFT + #undef BASE + #undef MASK + #ifdef SIZEOF_VOID_P + enum { __pyx_check_sizeof_voidp = 1 / (int)(SIZEOF_VOID_P == sizeof(void*)) }; + #endif +#endif +#ifndef __has_attribute + #define __has_attribute(x) 0 +#endif +#ifndef __has_cpp_attribute + #define __has_cpp_attribute(x) 0 +#endif +#ifndef CYTHON_RESTRICT + #if defined(__GNUC__) + #define CYTHON_RESTRICT __restrict__ + #elif defined(_MSC_VER) && _MSC_VER >= 1400 + #define CYTHON_RESTRICT __restrict + #elif defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L + #define CYTHON_RESTRICT restrict + #else + #define CYTHON_RESTRICT + #endif +#endif +#ifndef CYTHON_UNUSED +# if defined(__GNUC__) +# if !(defined(__cplusplus)) || (__GNUC__ > 3 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4)) +# define CYTHON_UNUSED __attribute__ ((__unused__)) +# else +# define CYTHON_UNUSED +# endif +# elif defined(__ICC) || (defined(__INTEL_COMPILER) && !defined(_MSC_VER)) +# define CYTHON_UNUSED __attribute__ ((__unused__)) +# else +# define CYTHON_UNUSED +# endif +#endif +#ifndef CYTHON_MAYBE_UNUSED_VAR +# if defined(__cplusplus) + template void CYTHON_MAYBE_UNUSED_VAR( const T& ) { } +# else +# define CYTHON_MAYBE_UNUSED_VAR(x) (void)(x) +# endif +#endif +#ifndef CYTHON_NCP_UNUSED +# if CYTHON_COMPILING_IN_CPYTHON +# define CYTHON_NCP_UNUSED +# else +# define CYTHON_NCP_UNUSED CYTHON_UNUSED +# endif +#endif +#define __Pyx_void_to_None(void_result) ((void)(void_result), Py_INCREF(Py_None), Py_None) +#ifdef _MSC_VER + #ifndef _MSC_STDINT_H_ + #if _MSC_VER < 1300 + typedef unsigned char uint8_t; + typedef unsigned int uint32_t; + #else + typedef unsigned __int8 uint8_t; + typedef unsigned __int32 uint32_t; + #endif + #endif +#else + #include +#endif +#ifndef CYTHON_FALLTHROUGH + #if defined(__cplusplus) && __cplusplus >= 201103L + #if __has_cpp_attribute(fallthrough) + #define CYTHON_FALLTHROUGH [[fallthrough]] + #elif __has_cpp_attribute(clang::fallthrough) + #define CYTHON_FALLTHROUGH [[clang::fallthrough]] + #elif __has_cpp_attribute(gnu::fallthrough) + #define CYTHON_FALLTHROUGH [[gnu::fallthrough]] + #endif + #endif + #ifndef CYTHON_FALLTHROUGH + #if __has_attribute(fallthrough) + #define CYTHON_FALLTHROUGH __attribute__((fallthrough)) + #else + #define CYTHON_FALLTHROUGH + #endif + #endif + #if defined(__clang__ ) && defined(__apple_build_version__) + #if __apple_build_version__ < 7000000 + #undef CYTHON_FALLTHROUGH + #define CYTHON_FALLTHROUGH + #endif + #endif +#endif + +#ifndef CYTHON_INLINE + #if defined(__clang__) + #define CYTHON_INLINE __inline__ __attribute__ ((__unused__)) + #elif defined(__GNUC__) + #define CYTHON_INLINE __inline__ + #elif defined(_MSC_VER) + #define CYTHON_INLINE __inline + #elif defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L + #define CYTHON_INLINE inline + #else + #define CYTHON_INLINE + #endif +#endif + +#if CYTHON_COMPILING_IN_PYPY && PY_VERSION_HEX < 0x02070600 && !defined(Py_OptimizeFlag) + #define Py_OptimizeFlag 0 +#endif +#define __PYX_BUILD_PY_SSIZE_T "n" +#define CYTHON_FORMAT_SSIZE_T "z" +#if PY_MAJOR_VERSION < 3 + #define __Pyx_BUILTIN_MODULE_NAME "__builtin__" + #define __Pyx_PyCode_New(a, k, l, s, f, code, c, n, v, fv, cell, fn, name, fline, lnos)\ + PyCode_New(a+k, l, s, f, code, c, n, v, fv, cell, fn, name, fline, lnos) + #define __Pyx_DefaultClassType PyClass_Type +#else + #define __Pyx_BUILTIN_MODULE_NAME "builtins" +#if PY_VERSION_HEX >= 0x030800A4 && PY_VERSION_HEX < 0x030800B2 + #define __Pyx_PyCode_New(a, k, l, s, f, code, c, n, v, fv, cell, fn, name, fline, lnos)\ + PyCode_New(a, 0, k, l, s, f, code, c, n, v, fv, cell, fn, name, fline, lnos) +#else + #define __Pyx_PyCode_New(a, k, l, s, f, code, c, n, v, fv, cell, fn, name, fline, lnos)\ + PyCode_New(a, k, l, s, f, code, c, n, v, fv, cell, fn, name, fline, lnos) +#endif + #define __Pyx_DefaultClassType PyType_Type +#endif +#ifndef Py_TPFLAGS_CHECKTYPES + #define Py_TPFLAGS_CHECKTYPES 0 +#endif +#ifndef Py_TPFLAGS_HAVE_INDEX + #define Py_TPFLAGS_HAVE_INDEX 0 +#endif +#ifndef Py_TPFLAGS_HAVE_NEWBUFFER + #define Py_TPFLAGS_HAVE_NEWBUFFER 0 +#endif +#ifndef Py_TPFLAGS_HAVE_FINALIZE + #define Py_TPFLAGS_HAVE_FINALIZE 0 +#endif +#ifndef METH_STACKLESS + #define METH_STACKLESS 0 +#endif +#if PY_VERSION_HEX <= 0x030700A3 || !defined(METH_FASTCALL) + #ifndef METH_FASTCALL + #define METH_FASTCALL 0x80 + #endif + typedef PyObject *(*__Pyx_PyCFunctionFast) (PyObject *self, PyObject *const *args, Py_ssize_t nargs); + typedef PyObject *(*__Pyx_PyCFunctionFastWithKeywords) (PyObject *self, PyObject *const *args, + Py_ssize_t nargs, PyObject *kwnames); +#else + #define __Pyx_PyCFunctionFast _PyCFunctionFast + #define __Pyx_PyCFunctionFastWithKeywords _PyCFunctionFastWithKeywords +#endif +#if CYTHON_FAST_PYCCALL +#define __Pyx_PyFastCFunction_Check(func)\ + ((PyCFunction_Check(func) && (METH_FASTCALL == (PyCFunction_GET_FLAGS(func) & ~(METH_CLASS | METH_STATIC | METH_COEXIST | METH_KEYWORDS | METH_STACKLESS))))) +#else +#define __Pyx_PyFastCFunction_Check(func) 0 +#endif +#if CYTHON_COMPILING_IN_PYPY && !defined(PyObject_Malloc) + #define PyObject_Malloc(s) PyMem_Malloc(s) + #define PyObject_Free(p) PyMem_Free(p) + #define PyObject_Realloc(p) PyMem_Realloc(p) +#endif +#if CYTHON_COMPILING_IN_CPYTHON && PY_VERSION_HEX < 0x030400A1 + #define PyMem_RawMalloc(n) PyMem_Malloc(n) + #define PyMem_RawRealloc(p, n) PyMem_Realloc(p, n) + #define PyMem_RawFree(p) PyMem_Free(p) +#endif +#if CYTHON_COMPILING_IN_PYSTON + #define __Pyx_PyCode_HasFreeVars(co) PyCode_HasFreeVars(co) + #define __Pyx_PyFrame_SetLineNumber(frame, lineno) PyFrame_SetLineNumber(frame, lineno) +#else + #define __Pyx_PyCode_HasFreeVars(co) (PyCode_GetNumFree(co) > 0) + #define __Pyx_PyFrame_SetLineNumber(frame, lineno) (frame)->f_lineno = (lineno) +#endif +#if !CYTHON_FAST_THREAD_STATE || PY_VERSION_HEX < 0x02070000 + #define __Pyx_PyThreadState_Current PyThreadState_GET() +#elif PY_VERSION_HEX >= 0x03060000 + #define __Pyx_PyThreadState_Current _PyThreadState_UncheckedGet() +#elif PY_VERSION_HEX >= 0x03000000 + #define __Pyx_PyThreadState_Current PyThreadState_GET() +#else + #define __Pyx_PyThreadState_Current _PyThreadState_Current +#endif +#if PY_VERSION_HEX < 0x030700A2 && !defined(PyThread_tss_create) && !defined(Py_tss_NEEDS_INIT) +#include "pythread.h" +#define Py_tss_NEEDS_INIT 0 +typedef int Py_tss_t; +static CYTHON_INLINE int PyThread_tss_create(Py_tss_t *key) { + *key = PyThread_create_key(); + return 0; +} +static CYTHON_INLINE Py_tss_t * PyThread_tss_alloc(void) { + Py_tss_t *key = (Py_tss_t *)PyObject_Malloc(sizeof(Py_tss_t)); + *key = Py_tss_NEEDS_INIT; + return key; +} +static CYTHON_INLINE void PyThread_tss_free(Py_tss_t *key) { + PyObject_Free(key); +} +static CYTHON_INLINE int PyThread_tss_is_created(Py_tss_t *key) { + return *key != Py_tss_NEEDS_INIT; +} +static CYTHON_INLINE void PyThread_tss_delete(Py_tss_t *key) { + PyThread_delete_key(*key); + *key = Py_tss_NEEDS_INIT; +} +static CYTHON_INLINE int PyThread_tss_set(Py_tss_t *key, void *value) { + return PyThread_set_key_value(*key, value); +} +static CYTHON_INLINE void * PyThread_tss_get(Py_tss_t *key) { + return PyThread_get_key_value(*key); +} +#endif +#if CYTHON_COMPILING_IN_CPYTHON || defined(_PyDict_NewPresized) +#define __Pyx_PyDict_NewPresized(n) ((n <= 8) ? PyDict_New() : _PyDict_NewPresized(n)) +#else +#define __Pyx_PyDict_NewPresized(n) PyDict_New() +#endif +#if PY_MAJOR_VERSION >= 3 || CYTHON_FUTURE_DIVISION + #define __Pyx_PyNumber_Divide(x,y) PyNumber_TrueDivide(x,y) + #define __Pyx_PyNumber_InPlaceDivide(x,y) PyNumber_InPlaceTrueDivide(x,y) +#else + #define __Pyx_PyNumber_Divide(x,y) PyNumber_Divide(x,y) + #define __Pyx_PyNumber_InPlaceDivide(x,y) PyNumber_InPlaceDivide(x,y) +#endif +#if CYTHON_COMPILING_IN_CPYTHON && PY_VERSION_HEX >= 0x030500A1 && CYTHON_USE_UNICODE_INTERNALS +#define __Pyx_PyDict_GetItemStr(dict, name) _PyDict_GetItem_KnownHash(dict, name, ((PyASCIIObject *) name)->hash) +#else +#define __Pyx_PyDict_GetItemStr(dict, name) PyDict_GetItem(dict, name) +#endif +#if PY_VERSION_HEX > 0x03030000 && defined(PyUnicode_KIND) + #define CYTHON_PEP393_ENABLED 1 + #define __Pyx_PyUnicode_READY(op) (likely(PyUnicode_IS_READY(op)) ?\ + 0 : _PyUnicode_Ready((PyObject *)(op))) + #define __Pyx_PyUnicode_GET_LENGTH(u) PyUnicode_GET_LENGTH(u) + #define __Pyx_PyUnicode_READ_CHAR(u, i) PyUnicode_READ_CHAR(u, i) + #define __Pyx_PyUnicode_MAX_CHAR_VALUE(u) PyUnicode_MAX_CHAR_VALUE(u) + #define __Pyx_PyUnicode_KIND(u) PyUnicode_KIND(u) + #define __Pyx_PyUnicode_DATA(u) PyUnicode_DATA(u) + #define __Pyx_PyUnicode_READ(k, d, i) PyUnicode_READ(k, d, i) + #define __Pyx_PyUnicode_WRITE(k, d, i, ch) PyUnicode_WRITE(k, d, i, ch) + #define __Pyx_PyUnicode_IS_TRUE(u) (0 != (likely(PyUnicode_IS_READY(u)) ? PyUnicode_GET_LENGTH(u) : PyUnicode_GET_SIZE(u))) +#else + #define CYTHON_PEP393_ENABLED 0 + #define PyUnicode_1BYTE_KIND 1 + #define PyUnicode_2BYTE_KIND 2 + #define PyUnicode_4BYTE_KIND 4 + #define __Pyx_PyUnicode_READY(op) (0) + #define __Pyx_PyUnicode_GET_LENGTH(u) PyUnicode_GET_SIZE(u) + #define __Pyx_PyUnicode_READ_CHAR(u, i) ((Py_UCS4)(PyUnicode_AS_UNICODE(u)[i])) + #define __Pyx_PyUnicode_MAX_CHAR_VALUE(u) ((sizeof(Py_UNICODE) == 2) ? 65535 : 1114111) + #define __Pyx_PyUnicode_KIND(u) (sizeof(Py_UNICODE)) + #define __Pyx_PyUnicode_DATA(u) ((void*)PyUnicode_AS_UNICODE(u)) + #define __Pyx_PyUnicode_READ(k, d, i) ((void)(k), (Py_UCS4)(((Py_UNICODE*)d)[i])) + #define __Pyx_PyUnicode_WRITE(k, d, i, ch) (((void)(k)), ((Py_UNICODE*)d)[i] = ch) + #define __Pyx_PyUnicode_IS_TRUE(u) (0 != PyUnicode_GET_SIZE(u)) +#endif +#if CYTHON_COMPILING_IN_PYPY + #define __Pyx_PyUnicode_Concat(a, b) PyNumber_Add(a, b) + #define __Pyx_PyUnicode_ConcatSafe(a, b) PyNumber_Add(a, b) +#else + #define __Pyx_PyUnicode_Concat(a, b) PyUnicode_Concat(a, b) + #define __Pyx_PyUnicode_ConcatSafe(a, b) ((unlikely((a) == Py_None) || unlikely((b) == Py_None)) ?\ + PyNumber_Add(a, b) : __Pyx_PyUnicode_Concat(a, b)) +#endif +#if CYTHON_COMPILING_IN_PYPY && !defined(PyUnicode_Contains) + #define PyUnicode_Contains(u, s) PySequence_Contains(u, s) +#endif +#if CYTHON_COMPILING_IN_PYPY && !defined(PyByteArray_Check) + #define PyByteArray_Check(obj) PyObject_TypeCheck(obj, &PyByteArray_Type) +#endif +#if CYTHON_COMPILING_IN_PYPY && !defined(PyObject_Format) + #define PyObject_Format(obj, fmt) PyObject_CallMethod(obj, "__format__", "O", fmt) +#endif +#define __Pyx_PyString_FormatSafe(a, b) ((unlikely((a) == Py_None || (PyString_Check(b) && !PyString_CheckExact(b)))) ? PyNumber_Remainder(a, b) : __Pyx_PyString_Format(a, b)) +#define __Pyx_PyUnicode_FormatSafe(a, b) ((unlikely((a) == Py_None || (PyUnicode_Check(b) && !PyUnicode_CheckExact(b)))) ? PyNumber_Remainder(a, b) : PyUnicode_Format(a, b)) +#if PY_MAJOR_VERSION >= 3 + #define __Pyx_PyString_Format(a, b) PyUnicode_Format(a, b) +#else + #define __Pyx_PyString_Format(a, b) PyString_Format(a, b) +#endif +#if PY_MAJOR_VERSION < 3 && !defined(PyObject_ASCII) + #define PyObject_ASCII(o) PyObject_Repr(o) +#endif +#if PY_MAJOR_VERSION >= 3 + #define PyBaseString_Type PyUnicode_Type + #define PyStringObject PyUnicodeObject + #define PyString_Type PyUnicode_Type + #define PyString_Check PyUnicode_Check + #define PyString_CheckExact PyUnicode_CheckExact + #define PyObject_Unicode PyObject_Str +#endif +#if PY_MAJOR_VERSION >= 3 + #define __Pyx_PyBaseString_Check(obj) PyUnicode_Check(obj) + #define __Pyx_PyBaseString_CheckExact(obj) PyUnicode_CheckExact(obj) +#else + #define __Pyx_PyBaseString_Check(obj) (PyString_Check(obj) || PyUnicode_Check(obj)) + #define __Pyx_PyBaseString_CheckExact(obj) (PyString_CheckExact(obj) || PyUnicode_CheckExact(obj)) +#endif +#ifndef PySet_CheckExact + #define PySet_CheckExact(obj) (Py_TYPE(obj) == &PySet_Type) +#endif +#if CYTHON_ASSUME_SAFE_MACROS + #define __Pyx_PySequence_SIZE(seq) Py_SIZE(seq) +#else + #define __Pyx_PySequence_SIZE(seq) PySequence_Size(seq) +#endif +#if PY_MAJOR_VERSION >= 3 + #define PyIntObject PyLongObject + #define PyInt_Type PyLong_Type + #define PyInt_Check(op) PyLong_Check(op) + #define PyInt_CheckExact(op) PyLong_CheckExact(op) + #define PyInt_FromString PyLong_FromString + #define PyInt_FromUnicode PyLong_FromUnicode + #define PyInt_FromLong PyLong_FromLong + #define PyInt_FromSize_t PyLong_FromSize_t + #define PyInt_FromSsize_t PyLong_FromSsize_t + #define PyInt_AsLong PyLong_AsLong + #define PyInt_AS_LONG PyLong_AS_LONG + #define PyInt_AsSsize_t PyLong_AsSsize_t + #define PyInt_AsUnsignedLongMask PyLong_AsUnsignedLongMask + #define PyInt_AsUnsignedLongLongMask PyLong_AsUnsignedLongLongMask + #define PyNumber_Int PyNumber_Long +#endif +#if PY_MAJOR_VERSION >= 3 + #define PyBoolObject PyLongObject +#endif +#if PY_MAJOR_VERSION >= 3 && CYTHON_COMPILING_IN_PYPY + #ifndef PyUnicode_InternFromString + #define PyUnicode_InternFromString(s) PyUnicode_FromString(s) + #endif +#endif +#if PY_VERSION_HEX < 0x030200A4 + typedef long Py_hash_t; + #define __Pyx_PyInt_FromHash_t PyInt_FromLong + #define __Pyx_PyInt_AsHash_t PyInt_AsLong +#else + #define __Pyx_PyInt_FromHash_t PyInt_FromSsize_t + #define __Pyx_PyInt_AsHash_t PyInt_AsSsize_t +#endif +#if PY_MAJOR_VERSION >= 3 + #define __Pyx_PyMethod_New(func, self, klass) ((self) ? PyMethod_New(func, self) : (Py_INCREF(func), func)) +#else + #define __Pyx_PyMethod_New(func, self, klass) PyMethod_New(func, self, klass) +#endif +#if CYTHON_USE_ASYNC_SLOTS + #if PY_VERSION_HEX >= 0x030500B1 + #define __Pyx_PyAsyncMethodsStruct PyAsyncMethods + #define __Pyx_PyType_AsAsync(obj) (Py_TYPE(obj)->tp_as_async) + #else + #define __Pyx_PyType_AsAsync(obj) ((__Pyx_PyAsyncMethodsStruct*) (Py_TYPE(obj)->tp_reserved)) + #endif +#else + #define __Pyx_PyType_AsAsync(obj) NULL +#endif +#ifndef __Pyx_PyAsyncMethodsStruct + typedef struct { + unaryfunc am_await; + unaryfunc am_aiter; + unaryfunc am_anext; + } __Pyx_PyAsyncMethodsStruct; +#endif + +#if defined(WIN32) || defined(MS_WINDOWS) + #define _USE_MATH_DEFINES +#endif +#include +#ifdef NAN +#define __PYX_NAN() ((float) NAN) +#else +static CYTHON_INLINE float __PYX_NAN() { + float value; + memset(&value, 0xFF, sizeof(value)); + return value; +} +#endif +#if defined(__CYGWIN__) && defined(_LDBL_EQ_DBL) +#define __Pyx_truncl trunc +#else +#define __Pyx_truncl truncl +#endif + + +#define __PYX_ERR(f_index, lineno, Ln_error) \ +{ \ + __pyx_filename = __pyx_f[f_index]; __pyx_lineno = lineno; __pyx_clineno = __LINE__; goto Ln_error; \ +} + +#ifndef __PYX_EXTERN_C + #ifdef __cplusplus + #define __PYX_EXTERN_C extern "C" + #else + #define __PYX_EXTERN_C extern + #endif +#endif + +#define __PYX_HAVE__average_inner +#define __PYX_HAVE_API__average_inner +/* Early includes */ +#include +#include +#include "numpy/arrayobject.h" +#include "numpy/ufuncobject.h" +#include "voidptr.h" +#ifdef _OPENMP +#include +#endif /* _OPENMP */ + +#if defined(PYREX_WITHOUT_ASSERTIONS) && !defined(CYTHON_WITHOUT_ASSERTIONS) +#define CYTHON_WITHOUT_ASSERTIONS +#endif + +typedef struct {PyObject **p; const char *s; const Py_ssize_t n; const char* encoding; + const char is_unicode; const char is_str; const char intern; } __Pyx_StringTabEntry; + +#define __PYX_DEFAULT_STRING_ENCODING_IS_ASCII 0 +#define __PYX_DEFAULT_STRING_ENCODING_IS_UTF8 0 +#define __PYX_DEFAULT_STRING_ENCODING_IS_DEFAULT (PY_MAJOR_VERSION >= 3 && __PYX_DEFAULT_STRING_ENCODING_IS_UTF8) +#define __PYX_DEFAULT_STRING_ENCODING "" +#define __Pyx_PyObject_FromString __Pyx_PyBytes_FromString +#define __Pyx_PyObject_FromStringAndSize __Pyx_PyBytes_FromStringAndSize +#define __Pyx_uchar_cast(c) ((unsigned char)c) +#define __Pyx_long_cast(x) ((long)x) +#define __Pyx_fits_Py_ssize_t(v, type, is_signed) (\ + (sizeof(type) < sizeof(Py_ssize_t)) ||\ + (sizeof(type) > sizeof(Py_ssize_t) &&\ + likely(v < (type)PY_SSIZE_T_MAX ||\ + v == (type)PY_SSIZE_T_MAX) &&\ + (!is_signed || likely(v > (type)PY_SSIZE_T_MIN ||\ + v == (type)PY_SSIZE_T_MIN))) ||\ + (sizeof(type) == sizeof(Py_ssize_t) &&\ + (is_signed || likely(v < (type)PY_SSIZE_T_MAX ||\ + v == (type)PY_SSIZE_T_MAX))) ) +static CYTHON_INLINE int __Pyx_is_valid_index(Py_ssize_t i, Py_ssize_t limit) { + return (size_t) i < (size_t) limit; +} +#if defined (__cplusplus) && __cplusplus >= 201103L + #include + #define __Pyx_sst_abs(value) std::abs(value) +#elif SIZEOF_INT >= SIZEOF_SIZE_T + #define __Pyx_sst_abs(value) abs(value) +#elif SIZEOF_LONG >= SIZEOF_SIZE_T + #define __Pyx_sst_abs(value) labs(value) +#elif defined (_MSC_VER) + #define __Pyx_sst_abs(value) ((Py_ssize_t)_abs64(value)) +#elif defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L + #define __Pyx_sst_abs(value) llabs(value) +#elif defined (__GNUC__) + #define __Pyx_sst_abs(value) __builtin_llabs(value) +#else + #define __Pyx_sst_abs(value) ((value<0) ? -value : value) +#endif +static CYTHON_INLINE const char* __Pyx_PyObject_AsString(PyObject*); +static CYTHON_INLINE const char* __Pyx_PyObject_AsStringAndSize(PyObject*, Py_ssize_t* length); +#define __Pyx_PyByteArray_FromString(s) PyByteArray_FromStringAndSize((const char*)s, strlen((const char*)s)) +#define __Pyx_PyByteArray_FromStringAndSize(s, l) PyByteArray_FromStringAndSize((const char*)s, l) +#define __Pyx_PyBytes_FromString PyBytes_FromString +#define __Pyx_PyBytes_FromStringAndSize PyBytes_FromStringAndSize +static CYTHON_INLINE PyObject* __Pyx_PyUnicode_FromString(const char*); +#if PY_MAJOR_VERSION < 3 + #define __Pyx_PyStr_FromString __Pyx_PyBytes_FromString + #define __Pyx_PyStr_FromStringAndSize __Pyx_PyBytes_FromStringAndSize +#else + #define __Pyx_PyStr_FromString __Pyx_PyUnicode_FromString + #define __Pyx_PyStr_FromStringAndSize __Pyx_PyUnicode_FromStringAndSize +#endif +#define __Pyx_PyBytes_AsWritableString(s) ((char*) PyBytes_AS_STRING(s)) +#define __Pyx_PyBytes_AsWritableSString(s) ((signed char*) PyBytes_AS_STRING(s)) +#define __Pyx_PyBytes_AsWritableUString(s) ((unsigned char*) PyBytes_AS_STRING(s)) +#define __Pyx_PyBytes_AsString(s) ((const char*) PyBytes_AS_STRING(s)) +#define __Pyx_PyBytes_AsSString(s) ((const signed char*) PyBytes_AS_STRING(s)) +#define __Pyx_PyBytes_AsUString(s) ((const unsigned char*) PyBytes_AS_STRING(s)) +#define __Pyx_PyObject_AsWritableString(s) ((char*) __Pyx_PyObject_AsString(s)) +#define __Pyx_PyObject_AsWritableSString(s) ((signed char*) __Pyx_PyObject_AsString(s)) +#define __Pyx_PyObject_AsWritableUString(s) ((unsigned char*) __Pyx_PyObject_AsString(s)) +#define __Pyx_PyObject_AsSString(s) ((const signed char*) __Pyx_PyObject_AsString(s)) +#define __Pyx_PyObject_AsUString(s) ((const unsigned char*) __Pyx_PyObject_AsString(s)) +#define __Pyx_PyObject_FromCString(s) __Pyx_PyObject_FromString((const char*)s) +#define __Pyx_PyBytes_FromCString(s) __Pyx_PyBytes_FromString((const char*)s) +#define __Pyx_PyByteArray_FromCString(s) __Pyx_PyByteArray_FromString((const char*)s) +#define __Pyx_PyStr_FromCString(s) __Pyx_PyStr_FromString((const char*)s) +#define __Pyx_PyUnicode_FromCString(s) __Pyx_PyUnicode_FromString((const char*)s) +static CYTHON_INLINE size_t __Pyx_Py_UNICODE_strlen(const Py_UNICODE *u) { + const Py_UNICODE *u_end = u; + while (*u_end++) ; + return (size_t)(u_end - u - 1); +} +#define __Pyx_PyUnicode_FromUnicode(u) PyUnicode_FromUnicode(u, __Pyx_Py_UNICODE_strlen(u)) +#define __Pyx_PyUnicode_FromUnicodeAndLength PyUnicode_FromUnicode +#define __Pyx_PyUnicode_AsUnicode PyUnicode_AsUnicode +#define __Pyx_NewRef(obj) (Py_INCREF(obj), obj) +#define __Pyx_Owned_Py_None(b) __Pyx_NewRef(Py_None) +static CYTHON_INLINE PyObject * __Pyx_PyBool_FromLong(long b); +static CYTHON_INLINE int __Pyx_PyObject_IsTrue(PyObject*); +static CYTHON_INLINE int __Pyx_PyObject_IsTrueAndDecref(PyObject*); +static CYTHON_INLINE PyObject* __Pyx_PyNumber_IntOrLong(PyObject* x); +#define __Pyx_PySequence_Tuple(obj)\ + (likely(PyTuple_CheckExact(obj)) ? __Pyx_NewRef(obj) : PySequence_Tuple(obj)) +static CYTHON_INLINE Py_ssize_t __Pyx_PyIndex_AsSsize_t(PyObject*); +static CYTHON_INLINE PyObject * __Pyx_PyInt_FromSize_t(size_t); +#if CYTHON_ASSUME_SAFE_MACROS +#define __pyx_PyFloat_AsDouble(x) (PyFloat_CheckExact(x) ? PyFloat_AS_DOUBLE(x) : PyFloat_AsDouble(x)) +#else +#define __pyx_PyFloat_AsDouble(x) PyFloat_AsDouble(x) +#endif +#define __pyx_PyFloat_AsFloat(x) ((float) __pyx_PyFloat_AsDouble(x)) +#if PY_MAJOR_VERSION >= 3 +#define __Pyx_PyNumber_Int(x) (PyLong_CheckExact(x) ? __Pyx_NewRef(x) : PyNumber_Long(x)) +#else +#define __Pyx_PyNumber_Int(x) (PyInt_CheckExact(x) ? __Pyx_NewRef(x) : PyNumber_Int(x)) +#endif +#define __Pyx_PyNumber_Float(x) (PyFloat_CheckExact(x) ? __Pyx_NewRef(x) : PyNumber_Float(x)) +#if PY_MAJOR_VERSION < 3 && __PYX_DEFAULT_STRING_ENCODING_IS_ASCII +static int __Pyx_sys_getdefaultencoding_not_ascii; +static int __Pyx_init_sys_getdefaultencoding_params(void) { + PyObject* sys; + PyObject* default_encoding = NULL; + PyObject* ascii_chars_u = NULL; + PyObject* ascii_chars_b = NULL; + const char* default_encoding_c; + sys = PyImport_ImportModule("sys"); + if (!sys) goto bad; + default_encoding = PyObject_CallMethod(sys, (char*) "getdefaultencoding", NULL); + Py_DECREF(sys); + if (!default_encoding) goto bad; + default_encoding_c = PyBytes_AsString(default_encoding); + if (!default_encoding_c) goto bad; + if (strcmp(default_encoding_c, "ascii") == 0) { + __Pyx_sys_getdefaultencoding_not_ascii = 0; + } else { + char ascii_chars[128]; + int c; + for (c = 0; c < 128; c++) { + ascii_chars[c] = c; + } + __Pyx_sys_getdefaultencoding_not_ascii = 1; + ascii_chars_u = PyUnicode_DecodeASCII(ascii_chars, 128, NULL); + if (!ascii_chars_u) goto bad; + ascii_chars_b = PyUnicode_AsEncodedString(ascii_chars_u, default_encoding_c, NULL); + if (!ascii_chars_b || !PyBytes_Check(ascii_chars_b) || memcmp(ascii_chars, PyBytes_AS_STRING(ascii_chars_b), 128) != 0) { + PyErr_Format( + PyExc_ValueError, + "This module compiled with c_string_encoding=ascii, but default encoding '%.200s' is not a superset of ascii.", + default_encoding_c); + goto bad; + } + Py_DECREF(ascii_chars_u); + Py_DECREF(ascii_chars_b); + } + Py_DECREF(default_encoding); + return 0; +bad: + Py_XDECREF(default_encoding); + Py_XDECREF(ascii_chars_u); + Py_XDECREF(ascii_chars_b); + return -1; +} +#endif +#if __PYX_DEFAULT_STRING_ENCODING_IS_DEFAULT && PY_MAJOR_VERSION >= 3 +#define __Pyx_PyUnicode_FromStringAndSize(c_str, size) PyUnicode_DecodeUTF8(c_str, size, NULL) +#else +#define __Pyx_PyUnicode_FromStringAndSize(c_str, size) PyUnicode_Decode(c_str, size, __PYX_DEFAULT_STRING_ENCODING, NULL) +#if __PYX_DEFAULT_STRING_ENCODING_IS_DEFAULT +static char* __PYX_DEFAULT_STRING_ENCODING; +static int __Pyx_init_sys_getdefaultencoding_params(void) { + PyObject* sys; + PyObject* default_encoding = NULL; + char* default_encoding_c; + sys = PyImport_ImportModule("sys"); + if (!sys) goto bad; + default_encoding = PyObject_CallMethod(sys, (char*) (const char*) "getdefaultencoding", NULL); + Py_DECREF(sys); + if (!default_encoding) goto bad; + default_encoding_c = PyBytes_AsString(default_encoding); + if (!default_encoding_c) goto bad; + __PYX_DEFAULT_STRING_ENCODING = (char*) malloc(strlen(default_encoding_c) + 1); + if (!__PYX_DEFAULT_STRING_ENCODING) goto bad; + strcpy(__PYX_DEFAULT_STRING_ENCODING, default_encoding_c); + Py_DECREF(default_encoding); + return 0; +bad: + Py_XDECREF(default_encoding); + return -1; +} +#endif +#endif + + +/* Test for GCC > 2.95 */ +#if defined(__GNUC__) && (__GNUC__ > 2 || (__GNUC__ == 2 && (__GNUC_MINOR__ > 95))) + #define likely(x) __builtin_expect(!!(x), 1) + #define unlikely(x) __builtin_expect(!!(x), 0) +#else /* !__GNUC__ or GCC < 2.95 */ + #define likely(x) (x) + #define unlikely(x) (x) +#endif /* __GNUC__ */ +static CYTHON_INLINE void __Pyx_pretend_to_initialize(void* ptr) { (void)ptr; } + +static PyObject *__pyx_m = NULL; +static PyObject *__pyx_d; +static PyObject *__pyx_b; +static PyObject *__pyx_cython_runtime = NULL; +static PyObject *__pyx_empty_tuple; +static PyObject *__pyx_empty_bytes; +static PyObject *__pyx_empty_unicode; +static int __pyx_lineno; +static int __pyx_clineno = 0; +static const char * __pyx_cfilenm= __FILE__; +static const char *__pyx_filename; + +/* Header.proto */ +#if !defined(CYTHON_CCOMPLEX) + #if defined(__cplusplus) + #define CYTHON_CCOMPLEX 1 + #elif defined(_Complex_I) + #define CYTHON_CCOMPLEX 1 + #else + #define CYTHON_CCOMPLEX 0 + #endif +#endif +#if CYTHON_CCOMPLEX + #ifdef __cplusplus + #include + #else + #include + #endif +#endif +#if CYTHON_CCOMPLEX && !defined(__cplusplus) && defined(__sun__) && defined(__GNUC__) + #undef _Complex_I + #define _Complex_I 1.0fj +#endif + + +static const char *__pyx_f[] = { + "average_inner.pyx", + "__init__.pxd", + "type.pxd", +}; +/* NoFastGil.proto */ +#define __Pyx_PyGILState_Ensure PyGILState_Ensure +#define __Pyx_PyGILState_Release PyGILState_Release +#define __Pyx_FastGIL_Remember() +#define __Pyx_FastGIL_Forget() +#define __Pyx_FastGilFuncInit() + +/* ForceInitThreads.proto */ +#ifndef __PYX_FORCE_INIT_THREADS + #define __PYX_FORCE_INIT_THREADS 0 +#endif + + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":776 + * # in Cython to enable them only on the right systems. + * + * ctypedef npy_int8 int8_t # <<<<<<<<<<<<<< + * ctypedef npy_int16 int16_t + * ctypedef npy_int32 int32_t + */ +typedef npy_int8 __pyx_t_5numpy_int8_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":777 + * + * ctypedef npy_int8 int8_t + * ctypedef npy_int16 int16_t # <<<<<<<<<<<<<< + * ctypedef npy_int32 int32_t + * ctypedef npy_int64 int64_t + */ +typedef npy_int16 __pyx_t_5numpy_int16_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":778 + * ctypedef npy_int8 int8_t + * ctypedef npy_int16 int16_t + * ctypedef npy_int32 int32_t # <<<<<<<<<<<<<< + * ctypedef npy_int64 int64_t + * #ctypedef npy_int96 int96_t + */ +typedef npy_int32 __pyx_t_5numpy_int32_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":779 + * ctypedef npy_int16 int16_t + * ctypedef npy_int32 int32_t + * ctypedef npy_int64 int64_t # <<<<<<<<<<<<<< + * #ctypedef npy_int96 int96_t + * #ctypedef npy_int128 int128_t + */ +typedef npy_int64 __pyx_t_5numpy_int64_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":783 + * #ctypedef npy_int128 int128_t + * + * ctypedef npy_uint8 uint8_t # <<<<<<<<<<<<<< + * ctypedef npy_uint16 uint16_t + * ctypedef npy_uint32 uint32_t + */ +typedef npy_uint8 __pyx_t_5numpy_uint8_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":784 + * + * ctypedef npy_uint8 uint8_t + * ctypedef npy_uint16 uint16_t # <<<<<<<<<<<<<< + * ctypedef npy_uint32 uint32_t + * ctypedef npy_uint64 uint64_t + */ +typedef npy_uint16 __pyx_t_5numpy_uint16_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":785 + * ctypedef npy_uint8 uint8_t + * ctypedef npy_uint16 uint16_t + * ctypedef npy_uint32 uint32_t # <<<<<<<<<<<<<< + * ctypedef npy_uint64 uint64_t + * #ctypedef npy_uint96 uint96_t + */ +typedef npy_uint32 __pyx_t_5numpy_uint32_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":786 + * ctypedef npy_uint16 uint16_t + * ctypedef npy_uint32 uint32_t + * ctypedef npy_uint64 uint64_t # <<<<<<<<<<<<<< + * #ctypedef npy_uint96 uint96_t + * #ctypedef npy_uint128 uint128_t + */ +typedef npy_uint64 __pyx_t_5numpy_uint64_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":790 + * #ctypedef npy_uint128 uint128_t + * + * ctypedef npy_float32 float32_t # <<<<<<<<<<<<<< + * ctypedef npy_float64 float64_t + * #ctypedef npy_float80 float80_t + */ +typedef npy_float32 __pyx_t_5numpy_float32_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":791 + * + * ctypedef npy_float32 float32_t + * ctypedef npy_float64 float64_t # <<<<<<<<<<<<<< + * #ctypedef npy_float80 float80_t + * #ctypedef npy_float128 float128_t + */ +typedef npy_float64 __pyx_t_5numpy_float64_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":800 + * # The int types are mapped a bit surprising -- + * # numpy.int corresponds to 'l' and numpy.long to 'q' + * ctypedef npy_long int_t # <<<<<<<<<<<<<< + * ctypedef npy_longlong long_t + * ctypedef npy_longlong longlong_t + */ +typedef npy_long __pyx_t_5numpy_int_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":801 + * # numpy.int corresponds to 'l' and numpy.long to 'q' + * ctypedef npy_long int_t + * ctypedef npy_longlong long_t # <<<<<<<<<<<<<< + * ctypedef npy_longlong longlong_t + * + */ +typedef npy_longlong __pyx_t_5numpy_long_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":802 + * ctypedef npy_long int_t + * ctypedef npy_longlong long_t + * ctypedef npy_longlong longlong_t # <<<<<<<<<<<<<< + * + * ctypedef npy_ulong uint_t + */ +typedef npy_longlong __pyx_t_5numpy_longlong_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":804 + * ctypedef npy_longlong longlong_t + * + * ctypedef npy_ulong uint_t # <<<<<<<<<<<<<< + * ctypedef npy_ulonglong ulong_t + * ctypedef npy_ulonglong ulonglong_t + */ +typedef npy_ulong __pyx_t_5numpy_uint_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":805 + * + * ctypedef npy_ulong uint_t + * ctypedef npy_ulonglong ulong_t # <<<<<<<<<<<<<< + * ctypedef npy_ulonglong ulonglong_t + * + */ +typedef npy_ulonglong __pyx_t_5numpy_ulong_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":806 + * ctypedef npy_ulong uint_t + * ctypedef npy_ulonglong ulong_t + * ctypedef npy_ulonglong ulonglong_t # <<<<<<<<<<<<<< + * + * ctypedef npy_intp intp_t + */ +typedef npy_ulonglong __pyx_t_5numpy_ulonglong_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":808 + * ctypedef npy_ulonglong ulonglong_t + * + * ctypedef npy_intp intp_t # <<<<<<<<<<<<<< + * ctypedef npy_uintp uintp_t + * + */ +typedef npy_intp __pyx_t_5numpy_intp_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":809 + * + * ctypedef npy_intp intp_t + * ctypedef npy_uintp uintp_t # <<<<<<<<<<<<<< + * + * ctypedef npy_double float_t + */ +typedef npy_uintp __pyx_t_5numpy_uintp_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":811 + * ctypedef npy_uintp uintp_t + * + * ctypedef npy_double float_t # <<<<<<<<<<<<<< + * ctypedef npy_double double_t + * ctypedef npy_longdouble longdouble_t + */ +typedef npy_double __pyx_t_5numpy_float_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":812 + * + * ctypedef npy_double float_t + * ctypedef npy_double double_t # <<<<<<<<<<<<<< + * ctypedef npy_longdouble longdouble_t + * + */ +typedef npy_double __pyx_t_5numpy_double_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":813 + * ctypedef npy_double float_t + * ctypedef npy_double double_t + * ctypedef npy_longdouble longdouble_t # <<<<<<<<<<<<<< + * + * ctypedef npy_cfloat cfloat_t + */ +typedef npy_longdouble __pyx_t_5numpy_longdouble_t; + +/* "average_inner.pxd":15 + * void* PyCObject_AsVoidPtr(object obj) + * + * ctypedef np.float32_t REAL_t # <<<<<<<<<<<<<< + * ctypedef np.uint32_t uINT_t + * + */ +typedef __pyx_t_5numpy_float32_t __pyx_t_13average_inner_REAL_t; + +/* "average_inner.pxd":16 + * + * ctypedef np.float32_t REAL_t + * ctypedef np.uint32_t uINT_t # <<<<<<<<<<<<<< + * + * # BLAS routine signatures + */ +typedef __pyx_t_5numpy_uint32_t __pyx_t_13average_inner_uINT_t; +/* Declarations.proto */ +#if CYTHON_CCOMPLEX + #ifdef __cplusplus + typedef ::std::complex< float > __pyx_t_float_complex; + #else + typedef float _Complex __pyx_t_float_complex; + #endif +#else + typedef struct { float real, imag; } __pyx_t_float_complex; +#endif +static CYTHON_INLINE __pyx_t_float_complex __pyx_t_float_complex_from_parts(float, float); + +/* Declarations.proto */ +#if CYTHON_CCOMPLEX + #ifdef __cplusplus + typedef ::std::complex< double > __pyx_t_double_complex; + #else + typedef double _Complex __pyx_t_double_complex; + #endif +#else + typedef struct { double real, imag; } __pyx_t_double_complex; +#endif +static CYTHON_INLINE __pyx_t_double_complex __pyx_t_double_complex_from_parts(double, double); + + +/*--- Type declarations ---*/ + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":815 + * ctypedef npy_longdouble longdouble_t + * + * ctypedef npy_cfloat cfloat_t # <<<<<<<<<<<<<< + * ctypedef npy_cdouble cdouble_t + * ctypedef npy_clongdouble clongdouble_t + */ +typedef npy_cfloat __pyx_t_5numpy_cfloat_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":816 + * + * ctypedef npy_cfloat cfloat_t + * ctypedef npy_cdouble cdouble_t # <<<<<<<<<<<<<< + * ctypedef npy_clongdouble clongdouble_t + * + */ +typedef npy_cdouble __pyx_t_5numpy_cdouble_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":817 + * ctypedef npy_cfloat cfloat_t + * ctypedef npy_cdouble cdouble_t + * ctypedef npy_clongdouble clongdouble_t # <<<<<<<<<<<<<< + * + * ctypedef npy_cdouble complex_t + */ +typedef npy_clongdouble __pyx_t_5numpy_clongdouble_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":819 + * ctypedef npy_clongdouble clongdouble_t + * + * ctypedef npy_cdouble complex_t # <<<<<<<<<<<<<< + * + * cdef inline object PyArray_MultiIterNew1(a): + */ +typedef npy_cdouble __pyx_t_5numpy_complex_t; +struct __pyx_t_13average_inner_BaseSentenceVecsConfig; +struct __pyx_t_13average_inner_FTSentenceVecsConfig; + +/* "average_inner.pxd":19 + * + * # BLAS routine signatures + * ctypedef void (*saxpy_ptr) (const int *N, const float *alpha, const float *X, const int *incX, float *Y, const int *incY) nogil # <<<<<<<<<<<<<< + * ctypedef void (*sscal_ptr) (const int *N, const float *alpha, const float *X, const int *incX) nogil + * + */ +typedef void (*__pyx_t_13average_inner_saxpy_ptr)(int const *, float const *, float const *, int const *, float *, int const *); + +/* "average_inner.pxd":20 + * # BLAS routine signatures + * ctypedef void (*saxpy_ptr) (const int *N, const float *alpha, const float *X, const int *incX, float *Y, const int *incY) nogil + * ctypedef void (*sscal_ptr) (const int *N, const float *alpha, const float *X, const int *incX) nogil # <<<<<<<<<<<<<< + * + * cdef saxpy_ptr saxpy + */ +typedef void (*__pyx_t_13average_inner_sscal_ptr)(int const *, float const *, float const *, int const *); + +/* "average_inner.pxd":34 + * DEF MAX_NGRAMS = 40 + * + * cdef struct BaseSentenceVecsConfig: # <<<<<<<<<<<<<< + * int size, workers + * + */ +struct __pyx_t_13average_inner_BaseSentenceVecsConfig { + int size; + int workers; + __pyx_t_13average_inner_REAL_t *mem; + __pyx_t_13average_inner_REAL_t *mem2; + __pyx_t_13average_inner_REAL_t *word_vectors; + __pyx_t_13average_inner_REAL_t *word_weights; + __pyx_t_13average_inner_REAL_t *sentence_vectors; + __pyx_t_13average_inner_uINT_t word_indices[0x2710]; + __pyx_t_13average_inner_uINT_t sent_adresses[0x2710]; + __pyx_t_13average_inner_uINT_t sentence_boundary[(0x2710 + 1)]; +}; + +/* "average_inner.pxd":48 + * uINT_t sentence_boundary[MAX_WORDS + 1] + * + * cdef struct FTSentenceVecsConfig: # <<<<<<<<<<<<<< + * int size, workers, min_n, max_n, bucket + * + */ +struct __pyx_t_13average_inner_FTSentenceVecsConfig { + int size; + int workers; + int min_n; + int max_n; + int bucket; + __pyx_t_13average_inner_REAL_t oov_weight; + __pyx_t_13average_inner_REAL_t *mem; + __pyx_t_13average_inner_REAL_t *mem2; + __pyx_t_13average_inner_REAL_t *word_vectors; + __pyx_t_13average_inner_REAL_t *ngram_vectors; + __pyx_t_13average_inner_REAL_t *word_weights; + __pyx_t_13average_inner_REAL_t *sentence_vectors; + __pyx_t_13average_inner_uINT_t word_indices[0x2710]; + __pyx_t_13average_inner_uINT_t sent_adresses[0x2710]; + __pyx_t_13average_inner_uINT_t sentence_boundary[(0x2710 + 1)]; + __pyx_t_13average_inner_uINT_t subwords_idx_len[0x2710]; + __pyx_t_13average_inner_uINT_t *subwords_idx; +}; + +/* --- Runtime support code (head) --- */ +/* Refnanny.proto */ +#ifndef CYTHON_REFNANNY + #define CYTHON_REFNANNY 0 +#endif +#if CYTHON_REFNANNY + typedef struct { + void (*INCREF)(void*, PyObject*, int); + void (*DECREF)(void*, PyObject*, int); + void (*GOTREF)(void*, PyObject*, int); + void (*GIVEREF)(void*, PyObject*, int); + void* (*SetupContext)(const char*, int, const char*); + void (*FinishContext)(void**); + } __Pyx_RefNannyAPIStruct; + static __Pyx_RefNannyAPIStruct *__Pyx_RefNanny = NULL; + static __Pyx_RefNannyAPIStruct *__Pyx_RefNannyImportAPI(const char *modname); + #define __Pyx_RefNannyDeclarations void *__pyx_refnanny = NULL; +#ifdef WITH_THREAD + #define __Pyx_RefNannySetupContext(name, acquire_gil)\ + if (acquire_gil) {\ + PyGILState_STATE __pyx_gilstate_save = PyGILState_Ensure();\ + __pyx_refnanny = __Pyx_RefNanny->SetupContext((name), __LINE__, __FILE__);\ + PyGILState_Release(__pyx_gilstate_save);\ + } else {\ + __pyx_refnanny = __Pyx_RefNanny->SetupContext((name), __LINE__, __FILE__);\ + } +#else + #define __Pyx_RefNannySetupContext(name, acquire_gil)\ + __pyx_refnanny = __Pyx_RefNanny->SetupContext((name), __LINE__, __FILE__) +#endif + #define __Pyx_RefNannyFinishContext()\ + __Pyx_RefNanny->FinishContext(&__pyx_refnanny) + #define __Pyx_INCREF(r) __Pyx_RefNanny->INCREF(__pyx_refnanny, (PyObject *)(r), __LINE__) + #define __Pyx_DECREF(r) __Pyx_RefNanny->DECREF(__pyx_refnanny, (PyObject *)(r), __LINE__) + #define __Pyx_GOTREF(r) __Pyx_RefNanny->GOTREF(__pyx_refnanny, (PyObject *)(r), __LINE__) + #define __Pyx_GIVEREF(r) __Pyx_RefNanny->GIVEREF(__pyx_refnanny, (PyObject *)(r), __LINE__) + #define __Pyx_XINCREF(r) do { if((r) != NULL) {__Pyx_INCREF(r); }} while(0) + #define __Pyx_XDECREF(r) do { if((r) != NULL) {__Pyx_DECREF(r); }} while(0) + #define __Pyx_XGOTREF(r) do { if((r) != NULL) {__Pyx_GOTREF(r); }} while(0) + #define __Pyx_XGIVEREF(r) do { if((r) != NULL) {__Pyx_GIVEREF(r);}} while(0) +#else + #define __Pyx_RefNannyDeclarations + #define __Pyx_RefNannySetupContext(name, acquire_gil) + #define __Pyx_RefNannyFinishContext() + #define __Pyx_INCREF(r) Py_INCREF(r) + #define __Pyx_DECREF(r) Py_DECREF(r) + #define __Pyx_GOTREF(r) + #define __Pyx_GIVEREF(r) + #define __Pyx_XINCREF(r) Py_XINCREF(r) + #define __Pyx_XDECREF(r) Py_XDECREF(r) + #define __Pyx_XGOTREF(r) + #define __Pyx_XGIVEREF(r) +#endif +#define __Pyx_XDECREF_SET(r, v) do {\ + PyObject *tmp = (PyObject *) r;\ + r = v; __Pyx_XDECREF(tmp);\ + } while (0) +#define __Pyx_DECREF_SET(r, v) do {\ + PyObject *tmp = (PyObject *) r;\ + r = v; __Pyx_DECREF(tmp);\ + } while (0) +#define __Pyx_CLEAR(r) do { PyObject* tmp = ((PyObject*)(r)); r = NULL; __Pyx_DECREF(tmp);} while(0) +#define __Pyx_XCLEAR(r) do { if((r) != NULL) {PyObject* tmp = ((PyObject*)(r)); r = NULL; __Pyx_DECREF(tmp);}} while(0) + +/* PyObjectGetAttrStr.proto */ +#if CYTHON_USE_TYPE_SLOTS +static CYTHON_INLINE PyObject* __Pyx_PyObject_GetAttrStr(PyObject* obj, PyObject* attr_name); +#else +#define __Pyx_PyObject_GetAttrStr(o,n) PyObject_GetAttr(o,n) +#endif + +/* GetBuiltinName.proto */ +static PyObject *__Pyx_GetBuiltinName(PyObject *name); + +/* GetItemInt.proto */ +#define __Pyx_GetItemInt(o, i, type, is_signed, to_py_func, is_list, wraparound, boundscheck)\ + (__Pyx_fits_Py_ssize_t(i, type, is_signed) ?\ + __Pyx_GetItemInt_Fast(o, (Py_ssize_t)i, is_list, wraparound, boundscheck) :\ + (is_list ? (PyErr_SetString(PyExc_IndexError, "list index out of range"), (PyObject*)NULL) :\ + __Pyx_GetItemInt_Generic(o, to_py_func(i)))) +#define __Pyx_GetItemInt_List(o, i, type, is_signed, to_py_func, is_list, wraparound, boundscheck)\ + (__Pyx_fits_Py_ssize_t(i, type, is_signed) ?\ + __Pyx_GetItemInt_List_Fast(o, (Py_ssize_t)i, wraparound, boundscheck) :\ + (PyErr_SetString(PyExc_IndexError, "list index out of range"), (PyObject*)NULL)) +static CYTHON_INLINE PyObject *__Pyx_GetItemInt_List_Fast(PyObject *o, Py_ssize_t i, + int wraparound, int boundscheck); +#define __Pyx_GetItemInt_Tuple(o, i, type, is_signed, to_py_func, is_list, wraparound, boundscheck)\ + (__Pyx_fits_Py_ssize_t(i, type, is_signed) ?\ + __Pyx_GetItemInt_Tuple_Fast(o, (Py_ssize_t)i, wraparound, boundscheck) :\ + (PyErr_SetString(PyExc_IndexError, "tuple index out of range"), (PyObject*)NULL)) +static CYTHON_INLINE PyObject *__Pyx_GetItemInt_Tuple_Fast(PyObject *o, Py_ssize_t i, + int wraparound, int boundscheck); +static PyObject *__Pyx_GetItemInt_Generic(PyObject *o, PyObject* j); +static CYTHON_INLINE PyObject *__Pyx_GetItemInt_Fast(PyObject *o, Py_ssize_t i, + int is_list, int wraparound, int boundscheck); + +/* ExtTypeTest.proto */ +static CYTHON_INLINE int __Pyx_TypeTest(PyObject *obj, PyTypeObject *type); + +/* PyDictVersioning.proto */ +#if CYTHON_USE_DICT_VERSIONS && CYTHON_USE_TYPE_SLOTS +#define __PYX_DICT_VERSION_INIT ((PY_UINT64_T) -1) +#define __PYX_GET_DICT_VERSION(dict) (((PyDictObject*)(dict))->ma_version_tag) +#define __PYX_UPDATE_DICT_CACHE(dict, value, cache_var, version_var)\ + (version_var) = __PYX_GET_DICT_VERSION(dict);\ + (cache_var) = (value); +#define __PYX_PY_DICT_LOOKUP_IF_MODIFIED(VAR, DICT, LOOKUP) {\ + static PY_UINT64_T __pyx_dict_version = 0;\ + static PyObject *__pyx_dict_cached_value = NULL;\ + if (likely(__PYX_GET_DICT_VERSION(DICT) == __pyx_dict_version)) {\ + (VAR) = __pyx_dict_cached_value;\ + } else {\ + (VAR) = __pyx_dict_cached_value = (LOOKUP);\ + __pyx_dict_version = __PYX_GET_DICT_VERSION(DICT);\ + }\ +} +static CYTHON_INLINE PY_UINT64_T __Pyx_get_tp_dict_version(PyObject *obj); +static CYTHON_INLINE PY_UINT64_T __Pyx_get_object_dict_version(PyObject *obj); +static CYTHON_INLINE int __Pyx_object_dict_version_matches(PyObject* obj, PY_UINT64_T tp_dict_version, PY_UINT64_T obj_dict_version); +#else +#define __PYX_GET_DICT_VERSION(dict) (0) +#define __PYX_UPDATE_DICT_CACHE(dict, value, cache_var, version_var) +#define __PYX_PY_DICT_LOOKUP_IF_MODIFIED(VAR, DICT, LOOKUP) (VAR) = (LOOKUP); +#endif + +/* GetModuleGlobalName.proto */ +#if CYTHON_USE_DICT_VERSIONS +#define __Pyx_GetModuleGlobalName(var, name) {\ + static PY_UINT64_T __pyx_dict_version = 0;\ + static PyObject *__pyx_dict_cached_value = NULL;\ + (var) = (likely(__pyx_dict_version == __PYX_GET_DICT_VERSION(__pyx_d))) ?\ + (likely(__pyx_dict_cached_value) ? __Pyx_NewRef(__pyx_dict_cached_value) : __Pyx_GetBuiltinName(name)) :\ + __Pyx__GetModuleGlobalName(name, &__pyx_dict_version, &__pyx_dict_cached_value);\ +} +#define __Pyx_GetModuleGlobalNameUncached(var, name) {\ + PY_UINT64_T __pyx_dict_version;\ + PyObject *__pyx_dict_cached_value;\ + (var) = __Pyx__GetModuleGlobalName(name, &__pyx_dict_version, &__pyx_dict_cached_value);\ +} +static PyObject *__Pyx__GetModuleGlobalName(PyObject *name, PY_UINT64_T *dict_version, PyObject **dict_cached_value); +#else +#define __Pyx_GetModuleGlobalName(var, name) (var) = __Pyx__GetModuleGlobalName(name) +#define __Pyx_GetModuleGlobalNameUncached(var, name) (var) = __Pyx__GetModuleGlobalName(name) +static CYTHON_INLINE PyObject *__Pyx__GetModuleGlobalName(PyObject *name); +#endif + +/* PyCFunctionFastCall.proto */ +#if CYTHON_FAST_PYCCALL +static CYTHON_INLINE PyObject *__Pyx_PyCFunction_FastCall(PyObject *func, PyObject **args, Py_ssize_t nargs); +#else +#define __Pyx_PyCFunction_FastCall(func, args, nargs) (assert(0), NULL) +#endif + +/* PyFunctionFastCall.proto */ +#if CYTHON_FAST_PYCALL +#define __Pyx_PyFunction_FastCall(func, args, nargs)\ + __Pyx_PyFunction_FastCallDict((func), (args), (nargs), NULL) +#if 1 || PY_VERSION_HEX < 0x030600B1 +static PyObject *__Pyx_PyFunction_FastCallDict(PyObject *func, PyObject **args, Py_ssize_t nargs, PyObject *kwargs); +#else +#define __Pyx_PyFunction_FastCallDict(func, args, nargs, kwargs) _PyFunction_FastCallDict(func, args, nargs, kwargs) +#endif +#define __Pyx_BUILD_ASSERT_EXPR(cond)\ + (sizeof(char [1 - 2*!(cond)]) - 1) +#ifndef Py_MEMBER_SIZE +#define Py_MEMBER_SIZE(type, member) sizeof(((type *)0)->member) +#endif + static size_t __pyx_pyframe_localsplus_offset = 0; + #include "frameobject.h" + #define __Pxy_PyFrame_Initialize_Offsets()\ + ((void)__Pyx_BUILD_ASSERT_EXPR(sizeof(PyFrameObject) == offsetof(PyFrameObject, f_localsplus) + Py_MEMBER_SIZE(PyFrameObject, f_localsplus)),\ + (void)(__pyx_pyframe_localsplus_offset = ((size_t)PyFrame_Type.tp_basicsize) - Py_MEMBER_SIZE(PyFrameObject, f_localsplus))) + #define __Pyx_PyFrame_GetLocalsplus(frame)\ + (assert(__pyx_pyframe_localsplus_offset), (PyObject **)(((char *)(frame)) + __pyx_pyframe_localsplus_offset)) +#endif + +/* PyObjectCall.proto */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE PyObject* __Pyx_PyObject_Call(PyObject *func, PyObject *arg, PyObject *kw); +#else +#define __Pyx_PyObject_Call(func, arg, kw) PyObject_Call(func, arg, kw) +#endif + +/* PyObjectCall2Args.proto */ +static CYTHON_UNUSED PyObject* __Pyx_PyObject_Call2Args(PyObject* function, PyObject* arg1, PyObject* arg2); + +/* PyObjectCallMethO.proto */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallMethO(PyObject *func, PyObject *arg); +#endif + +/* PyObjectCallOneArg.proto */ +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallOneArg(PyObject *func, PyObject *arg); + +/* PySequenceContains.proto */ +static CYTHON_INLINE int __Pyx_PySequence_ContainsTF(PyObject* item, PyObject* seq, int eq) { + int result = PySequence_Contains(seq, item); + return unlikely(result < 0) ? result : (result == (eq == Py_EQ)); +} + +/* ObjectGetItem.proto */ +#if CYTHON_USE_TYPE_SLOTS +static CYTHON_INLINE PyObject *__Pyx_PyObject_GetItem(PyObject *obj, PyObject* key); +#else +#define __Pyx_PyObject_GetItem(obj, key) PyObject_GetItem(obj, key) +#endif + +/* ListCompAppend.proto */ +#if CYTHON_USE_PYLIST_INTERNALS && CYTHON_ASSUME_SAFE_MACROS +static CYTHON_INLINE int __Pyx_ListComp_Append(PyObject* list, PyObject* x) { + PyListObject* L = (PyListObject*) list; + Py_ssize_t len = Py_SIZE(list); + if (likely(L->allocated > len)) { + Py_INCREF(x); + PyList_SET_ITEM(list, len, x); + Py_SIZE(list) = len+1; + return 0; + } + return PyList_Append(list, x); +} +#else +#define __Pyx_ListComp_Append(L,x) PyList_Append(L,x) +#endif + +/* SliceTupleAndList.proto */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE PyObject* __Pyx_PyList_GetSlice(PyObject* src, Py_ssize_t start, Py_ssize_t stop); +static CYTHON_INLINE PyObject* __Pyx_PyTuple_GetSlice(PyObject* src, Py_ssize_t start, Py_ssize_t stop); +#else +#define __Pyx_PyList_GetSlice(seq, start, stop) PySequence_GetSlice(seq, start, stop) +#define __Pyx_PyTuple_GetSlice(seq, start, stop) PySequence_GetSlice(seq, start, stop) +#endif + +/* PyIntBinop.proto */ +#if !CYTHON_COMPILING_IN_PYPY +static PyObject* __Pyx_PyInt_AddObjC(PyObject *op1, PyObject *op2, long intval, int inplace, int zerodivision_check); +#else +#define __Pyx_PyInt_AddObjC(op1, op2, intval, inplace, zerodivision_check)\ + (inplace ? PyNumber_InPlaceAdd(op1, op2) : PyNumber_Add(op1, op2)) +#endif + +/* RaiseArgTupleInvalid.proto */ +static void __Pyx_RaiseArgtupleInvalid(const char* func_name, int exact, + Py_ssize_t num_min, Py_ssize_t num_max, Py_ssize_t num_found); + +/* RaiseDoubleKeywords.proto */ +static void __Pyx_RaiseDoubleKeywordsError(const char* func_name, PyObject* kw_name); + +/* ParseKeywords.proto */ +static int __Pyx_ParseOptionalKeywords(PyObject *kwds, PyObject **argnames[],\ + PyObject *kwds2, PyObject *values[], Py_ssize_t num_pos_args,\ + const char* function_name); + +/* RaiseTooManyValuesToUnpack.proto */ +static CYTHON_INLINE void __Pyx_RaiseTooManyValuesError(Py_ssize_t expected); + +/* RaiseNeedMoreValuesToUnpack.proto */ +static CYTHON_INLINE void __Pyx_RaiseNeedMoreValuesError(Py_ssize_t index); + +/* IterFinish.proto */ +static CYTHON_INLINE int __Pyx_IterFinish(void); + +/* UnpackItemEndCheck.proto */ +static int __Pyx_IternextUnpackEndCheck(PyObject *retval, Py_ssize_t expected); + +/* PyThreadStateGet.proto */ +#if CYTHON_FAST_THREAD_STATE +#define __Pyx_PyThreadState_declare PyThreadState *__pyx_tstate; +#define __Pyx_PyThreadState_assign __pyx_tstate = __Pyx_PyThreadState_Current; +#define __Pyx_PyErr_Occurred() __pyx_tstate->curexc_type +#else +#define __Pyx_PyThreadState_declare +#define __Pyx_PyThreadState_assign +#define __Pyx_PyErr_Occurred() PyErr_Occurred() +#endif + +/* PyErrFetchRestore.proto */ +#if CYTHON_FAST_THREAD_STATE +#define __Pyx_PyErr_Clear() __Pyx_ErrRestore(NULL, NULL, NULL) +#define __Pyx_ErrRestoreWithState(type, value, tb) __Pyx_ErrRestoreInState(PyThreadState_GET(), type, value, tb) +#define __Pyx_ErrFetchWithState(type, value, tb) __Pyx_ErrFetchInState(PyThreadState_GET(), type, value, tb) +#define __Pyx_ErrRestore(type, value, tb) __Pyx_ErrRestoreInState(__pyx_tstate, type, value, tb) +#define __Pyx_ErrFetch(type, value, tb) __Pyx_ErrFetchInState(__pyx_tstate, type, value, tb) +static CYTHON_INLINE void __Pyx_ErrRestoreInState(PyThreadState *tstate, PyObject *type, PyObject *value, PyObject *tb); +static CYTHON_INLINE void __Pyx_ErrFetchInState(PyThreadState *tstate, PyObject **type, PyObject **value, PyObject **tb); +#if CYTHON_COMPILING_IN_CPYTHON +#define __Pyx_PyErr_SetNone(exc) (Py_INCREF(exc), __Pyx_ErrRestore((exc), NULL, NULL)) +#else +#define __Pyx_PyErr_SetNone(exc) PyErr_SetNone(exc) +#endif +#else +#define __Pyx_PyErr_Clear() PyErr_Clear() +#define __Pyx_PyErr_SetNone(exc) PyErr_SetNone(exc) +#define __Pyx_ErrRestoreWithState(type, value, tb) PyErr_Restore(type, value, tb) +#define __Pyx_ErrFetchWithState(type, value, tb) PyErr_Fetch(type, value, tb) +#define __Pyx_ErrRestoreInState(tstate, type, value, tb) PyErr_Restore(type, value, tb) +#define __Pyx_ErrFetchInState(tstate, type, value, tb) PyErr_Fetch(type, value, tb) +#define __Pyx_ErrRestore(type, value, tb) PyErr_Restore(type, value, tb) +#define __Pyx_ErrFetch(type, value, tb) PyErr_Fetch(type, value, tb) +#endif + +/* RaiseException.proto */ +static void __Pyx_Raise(PyObject *type, PyObject *value, PyObject *tb, PyObject *cause); + +/* DictGetItem.proto */ +#if PY_MAJOR_VERSION >= 3 && !CYTHON_COMPILING_IN_PYPY +static PyObject *__Pyx_PyDict_GetItem(PyObject *d, PyObject* key); +#define __Pyx_PyObject_Dict_GetItem(obj, name)\ + (likely(PyDict_CheckExact(obj)) ?\ + __Pyx_PyDict_GetItem(obj, name) : PyObject_GetItem(obj, name)) +#else +#define __Pyx_PyDict_GetItem(d, key) PyObject_GetItem(d, key) +#define __Pyx_PyObject_Dict_GetItem(obj, name) PyObject_GetItem(obj, name) +#endif + +/* RaiseNoneIterError.proto */ +static CYTHON_INLINE void __Pyx_RaiseNoneNotIterableError(void); + +/* GetTopmostException.proto */ +#if CYTHON_USE_EXC_INFO_STACK +static _PyErr_StackItem * __Pyx_PyErr_GetTopmostException(PyThreadState *tstate); +#endif + +/* SaveResetException.proto */ +#if CYTHON_FAST_THREAD_STATE +#define __Pyx_ExceptionSave(type, value, tb) __Pyx__ExceptionSave(__pyx_tstate, type, value, tb) +static CYTHON_INLINE void __Pyx__ExceptionSave(PyThreadState *tstate, PyObject **type, PyObject **value, PyObject **tb); +#define __Pyx_ExceptionReset(type, value, tb) __Pyx__ExceptionReset(__pyx_tstate, type, value, tb) +static CYTHON_INLINE void __Pyx__ExceptionReset(PyThreadState *tstate, PyObject *type, PyObject *value, PyObject *tb); +#else +#define __Pyx_ExceptionSave(type, value, tb) PyErr_GetExcInfo(type, value, tb) +#define __Pyx_ExceptionReset(type, value, tb) PyErr_SetExcInfo(type, value, tb) +#endif + +/* PyErrExceptionMatches.proto */ +#if CYTHON_FAST_THREAD_STATE +#define __Pyx_PyErr_ExceptionMatches(err) __Pyx_PyErr_ExceptionMatchesInState(__pyx_tstate, err) +static CYTHON_INLINE int __Pyx_PyErr_ExceptionMatchesInState(PyThreadState* tstate, PyObject* err); +#else +#define __Pyx_PyErr_ExceptionMatches(err) PyErr_ExceptionMatches(err) +#endif + +/* GetException.proto */ +#if CYTHON_FAST_THREAD_STATE +#define __Pyx_GetException(type, value, tb) __Pyx__GetException(__pyx_tstate, type, value, tb) +static int __Pyx__GetException(PyThreadState *tstate, PyObject **type, PyObject **value, PyObject **tb); +#else +static int __Pyx_GetException(PyObject **type, PyObject **value, PyObject **tb); +#endif + +/* TypeImport.proto */ +#ifndef __PYX_HAVE_RT_ImportType_proto +#define __PYX_HAVE_RT_ImportType_proto +enum __Pyx_ImportType_CheckSize { + __Pyx_ImportType_CheckSize_Error = 0, + __Pyx_ImportType_CheckSize_Warn = 1, + __Pyx_ImportType_CheckSize_Ignore = 2 +}; +static PyTypeObject *__Pyx_ImportType(PyObject* module, const char *module_name, const char *class_name, size_t size, enum __Pyx_ImportType_CheckSize check_size); +#endif + +/* Import.proto */ +static PyObject *__Pyx_Import(PyObject *name, PyObject *from_list, int level); + +/* ImportFrom.proto */ +static PyObject* __Pyx_ImportFrom(PyObject* module, PyObject* name); + +/* PyObjectCallNoArg.proto */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallNoArg(PyObject *func); +#else +#define __Pyx_PyObject_CallNoArg(func) __Pyx_PyObject_Call(func, __pyx_empty_tuple, NULL) +#endif + +/* CLineInTraceback.proto */ +#ifdef CYTHON_CLINE_IN_TRACEBACK +#define __Pyx_CLineForTraceback(tstate, c_line) (((CYTHON_CLINE_IN_TRACEBACK)) ? c_line : 0) +#else +static int __Pyx_CLineForTraceback(PyThreadState *tstate, int c_line); +#endif + +/* CodeObjectCache.proto */ +typedef struct { + PyCodeObject* code_object; + int code_line; +} __Pyx_CodeObjectCacheEntry; +struct __Pyx_CodeObjectCache { + int count; + int max_count; + __Pyx_CodeObjectCacheEntry* entries; +}; +static struct __Pyx_CodeObjectCache __pyx_code_cache = {0,0,NULL}; +static int __pyx_bisect_code_objects(__Pyx_CodeObjectCacheEntry* entries, int count, int code_line); +static PyCodeObject *__pyx_find_code_object(int code_line); +static void __pyx_insert_code_object(int code_line, PyCodeObject* code_object); + +/* AddTraceback.proto */ +static void __Pyx_AddTraceback(const char *funcname, int c_line, + int py_line, const char *filename); + +/* CIntToPy.proto */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_long(long value); + +/* CIntToPy.proto */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_int(int value); + +/* CIntToPy.proto */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_npy_uint32(npy_uint32 value); + +/* RealImag.proto */ +#if CYTHON_CCOMPLEX + #ifdef __cplusplus + #define __Pyx_CREAL(z) ((z).real()) + #define __Pyx_CIMAG(z) ((z).imag()) + #else + #define __Pyx_CREAL(z) (__real__(z)) + #define __Pyx_CIMAG(z) (__imag__(z)) + #endif +#else + #define __Pyx_CREAL(z) ((z).real) + #define __Pyx_CIMAG(z) ((z).imag) +#endif +#if defined(__cplusplus) && CYTHON_CCOMPLEX\ + && (defined(_WIN32) || defined(__clang__) || (defined(__GNUC__) && (__GNUC__ >= 5 || __GNUC__ == 4 && __GNUC_MINOR__ >= 4 )) || __cplusplus >= 201103) + #define __Pyx_SET_CREAL(z,x) ((z).real(x)) + #define __Pyx_SET_CIMAG(z,y) ((z).imag(y)) +#else + #define __Pyx_SET_CREAL(z,x) __Pyx_CREAL(z) = (x) + #define __Pyx_SET_CIMAG(z,y) __Pyx_CIMAG(z) = (y) +#endif + +/* Arithmetic.proto */ +#if CYTHON_CCOMPLEX + #define __Pyx_c_eq_float(a, b) ((a)==(b)) + #define __Pyx_c_sum_float(a, b) ((a)+(b)) + #define __Pyx_c_diff_float(a, b) ((a)-(b)) + #define __Pyx_c_prod_float(a, b) ((a)*(b)) + #define __Pyx_c_quot_float(a, b) ((a)/(b)) + #define __Pyx_c_neg_float(a) (-(a)) + #ifdef __cplusplus + #define __Pyx_c_is_zero_float(z) ((z)==(float)0) + #define __Pyx_c_conj_float(z) (::std::conj(z)) + #if 1 + #define __Pyx_c_abs_float(z) (::std::abs(z)) + #define __Pyx_c_pow_float(a, b) (::std::pow(a, b)) + #endif + #else + #define __Pyx_c_is_zero_float(z) ((z)==0) + #define __Pyx_c_conj_float(z) (conjf(z)) + #if 1 + #define __Pyx_c_abs_float(z) (cabsf(z)) + #define __Pyx_c_pow_float(a, b) (cpowf(a, b)) + #endif + #endif +#else + static CYTHON_INLINE int __Pyx_c_eq_float(__pyx_t_float_complex, __pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_sum_float(__pyx_t_float_complex, __pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_diff_float(__pyx_t_float_complex, __pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_prod_float(__pyx_t_float_complex, __pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_quot_float(__pyx_t_float_complex, __pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_neg_float(__pyx_t_float_complex); + static CYTHON_INLINE int __Pyx_c_is_zero_float(__pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_conj_float(__pyx_t_float_complex); + #if 1 + static CYTHON_INLINE float __Pyx_c_abs_float(__pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_pow_float(__pyx_t_float_complex, __pyx_t_float_complex); + #endif +#endif + +/* Arithmetic.proto */ +#if CYTHON_CCOMPLEX + #define __Pyx_c_eq_double(a, b) ((a)==(b)) + #define __Pyx_c_sum_double(a, b) ((a)+(b)) + #define __Pyx_c_diff_double(a, b) ((a)-(b)) + #define __Pyx_c_prod_double(a, b) ((a)*(b)) + #define __Pyx_c_quot_double(a, b) ((a)/(b)) + #define __Pyx_c_neg_double(a) (-(a)) + #ifdef __cplusplus + #define __Pyx_c_is_zero_double(z) ((z)==(double)0) + #define __Pyx_c_conj_double(z) (::std::conj(z)) + #if 1 + #define __Pyx_c_abs_double(z) (::std::abs(z)) + #define __Pyx_c_pow_double(a, b) (::std::pow(a, b)) + #endif + #else + #define __Pyx_c_is_zero_double(z) ((z)==0) + #define __Pyx_c_conj_double(z) (conj(z)) + #if 1 + #define __Pyx_c_abs_double(z) (cabs(z)) + #define __Pyx_c_pow_double(a, b) (cpow(a, b)) + #endif + #endif +#else + static CYTHON_INLINE int __Pyx_c_eq_double(__pyx_t_double_complex, __pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_sum_double(__pyx_t_double_complex, __pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_diff_double(__pyx_t_double_complex, __pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_prod_double(__pyx_t_double_complex, __pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_quot_double(__pyx_t_double_complex, __pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_neg_double(__pyx_t_double_complex); + static CYTHON_INLINE int __Pyx_c_is_zero_double(__pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_conj_double(__pyx_t_double_complex); + #if 1 + static CYTHON_INLINE double __Pyx_c_abs_double(__pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_pow_double(__pyx_t_double_complex, __pyx_t_double_complex); + #endif +#endif + +/* CIntToPy.proto */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_enum__NPY_TYPES(enum NPY_TYPES value); + +/* CIntFromPy.proto */ +static CYTHON_INLINE int __Pyx_PyInt_As_int(PyObject *); + +/* CIntFromPy.proto */ +static CYTHON_INLINE npy_uint32 __Pyx_PyInt_As_npy_uint32(PyObject *); + +/* CIntFromPy.proto */ +static CYTHON_INLINE long __Pyx_PyInt_As_long(PyObject *); + +/* FastTypeChecks.proto */ +#if CYTHON_COMPILING_IN_CPYTHON +#define __Pyx_TypeCheck(obj, type) __Pyx_IsSubtype(Py_TYPE(obj), (PyTypeObject *)type) +static CYTHON_INLINE int __Pyx_IsSubtype(PyTypeObject *a, PyTypeObject *b); +static CYTHON_INLINE int __Pyx_PyErr_GivenExceptionMatches(PyObject *err, PyObject *type); +static CYTHON_INLINE int __Pyx_PyErr_GivenExceptionMatches2(PyObject *err, PyObject *type1, PyObject *type2); +#else +#define __Pyx_TypeCheck(obj, type) PyObject_TypeCheck(obj, (PyTypeObject *)type) +#define __Pyx_PyErr_GivenExceptionMatches(err, type) PyErr_GivenExceptionMatches(err, type) +#define __Pyx_PyErr_GivenExceptionMatches2(err, type1, type2) (PyErr_GivenExceptionMatches(err, type1) || PyErr_GivenExceptionMatches(err, type2)) +#endif +#define __Pyx_PyException_Check(obj) __Pyx_TypeCheck(obj, PyExc_Exception) + +/* CheckBinaryVersion.proto */ +static int __Pyx_check_binary_version(void); + +/* PyObjectSetAttrStr.proto */ +#if CYTHON_USE_TYPE_SLOTS +#define __Pyx_PyObject_DelAttrStr(o,n) __Pyx_PyObject_SetAttrStr(o, n, NULL) +static CYTHON_INLINE int __Pyx_PyObject_SetAttrStr(PyObject* obj, PyObject* attr_name, PyObject* value); +#else +#define __Pyx_PyObject_DelAttrStr(o,n) PyObject_DelAttr(o,n) +#define __Pyx_PyObject_SetAttrStr(o,n,v) PyObject_SetAttr(o,n,v) +#endif + +/* VoidPtrExport.proto */ +static int __Pyx_ExportVoidPtr(PyObject *name, void *p, const char *sig); + +/* FunctionExport.proto */ +static int __Pyx_ExportFunction(const char *name, void (*f)(void), const char *sig); + +/* InitStrings.proto */ +static int __Pyx_InitStrings(__Pyx_StringTabEntry *t); + + +/* Module declarations from 'cpython.buffer' */ + +/* Module declarations from 'libc.string' */ + +/* Module declarations from 'libc.stdio' */ + +/* Module declarations from '__builtin__' */ + +/* Module declarations from 'cpython.type' */ +static PyTypeObject *__pyx_ptype_7cpython_4type_type = 0; + +/* Module declarations from 'cpython' */ + +/* Module declarations from 'cpython.object' */ + +/* Module declarations from 'cpython.ref' */ + +/* Module declarations from 'cpython.mem' */ + +/* Module declarations from 'numpy' */ + +/* Module declarations from 'numpy' */ +static PyTypeObject *__pyx_ptype_5numpy_dtype = 0; +static PyTypeObject *__pyx_ptype_5numpy_flatiter = 0; +static PyTypeObject *__pyx_ptype_5numpy_broadcast = 0; +static PyTypeObject *__pyx_ptype_5numpy_ndarray = 0; +static PyTypeObject *__pyx_ptype_5numpy_ufunc = 0; +static CYTHON_INLINE char *__pyx_f_5numpy__util_dtypestring(PyArray_Descr *, char *, char *, int *); /*proto*/ + +/* Module declarations from 'cython' */ + +/* Module declarations from 'average_inner' */ +static __pyx_t_13average_inner_saxpy_ptr __pyx_v_13average_inner_saxpy; +static __pyx_t_13average_inner_sscal_ptr __pyx_v_13average_inner_sscal; +static int __pyx_v_13average_inner_ONE; +static int __pyx_v_13average_inner_ZERO; +static __pyx_t_13average_inner_REAL_t __pyx_v_13average_inner_ONEF; +static __pyx_t_13average_inner_REAL_t __pyx_v_13average_inner_ZEROF; +static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *, PyObject *); /*proto*/ +static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *, PyObject *); /*proto*/ +static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *); /*proto*/ +static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *); /*proto*/ +static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ +static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ +#define __Pyx_MODULE_NAME "average_inner" +extern int __pyx_module_is_main_average_inner; +int __pyx_module_is_main_average_inner = 0; + +/* Implementation of 'average_inner' */ +static PyObject *__pyx_builtin_enumerate; +static PyObject *__pyx_builtin_range; +static PyObject *__pyx_builtin_ValueError; +static PyObject *__pyx_builtin_RuntimeError; +static PyObject *__pyx_builtin_ImportError; +static const char __pyx_k__8[] = "*"; +static const char __pyx_k_ft[] = "ft"; +static const char __pyx_k_np[] = "np"; +static const char __pyx_k_sv[] = "sv"; +static const char __pyx_k_wv[] = "wv"; +static const char __pyx_k_ONE[] = "ONE"; +static const char __pyx_k_max[] = "max"; +static const char __pyx_k_w2v[] = "w2v"; +static const char __pyx_k_ONEF[] = "ONEF"; +static const char __pyx_k_ZERO[] = "ZERO"; +static const char __pyx_k_fill[] = "fill"; +static const char __pyx_k_init[] = "init"; +static const char __pyx_k_main[] = "__main__"; +static const char __pyx_k_name[] = "__name__"; +static const char __pyx_k_test[] = "__test__"; +static const char __pyx_k_ZEROF[] = "ZEROF"; +static const char __pyx_k_fblas[] = "fblas"; +static const char __pyx_k_index[] = "index"; +static const char __pyx_k_is_ft[] = "is_ft"; +static const char __pyx_k_max_n[] = "max_n"; +static const char __pyx_k_min_n[] = "min_n"; +static const char __pyx_k_model[] = "model"; +static const char __pyx_k_numpy[] = "numpy"; +static const char __pyx_k_range[] = "range"; +static const char __pyx_k_saxpy[] = "saxpy"; +static const char __pyx_k_sscal[] = "sscal"; +static const char __pyx_k_vocab[] = "vocab"; +static const char __pyx_k_bucket[] = "bucket"; +static const char __pyx_k_import[] = "__import__"; +static const char __pyx_k_memory[] = "memory"; +static const char __pyx_k_target[] = "target"; +static const char __pyx_k_vectors[] = "vectors"; +static const char __pyx_k_workers[] = "workers"; +static const char __pyx_k_cpointer[] = "_cpointer"; +static const char __pyx_k_pyx_capi[] = "__pyx_capi__"; +static const char __pyx_k_eff_words[] = "eff_words"; +static const char __pyx_k_enumerate[] = "enumerate"; +static const char __pyx_k_ValueError[] = "ValueError"; +static const char __pyx_k_ImportError[] = "ImportError"; +static const char __pyx_k_vector_size[] = "vector_size"; +static const char __pyx_k_FAST_VERSION[] = "FAST_VERSION"; +static const char __pyx_k_RuntimeError[] = "RuntimeError"; +static const char __pyx_k_word_weights[] = "word_weights"; +static const char __pyx_k_average_inner[] = "average_inner"; +static const char __pyx_k_eff_sentences[] = "eff_sentences"; +static const char __pyx_k_ft_hash_bytes[] = "ft_hash_bytes"; +static const char __pyx_k_vectors_vocab[] = "vectors_vocab"; +static const char __pyx_k_vectors_ngrams[] = "vectors_ngrams"; +static const char __pyx_k_train_average_cy[] = "train_average_cy"; +static const char __pyx_k_average_inner_pyx[] = "average_inner.pyx"; +static const char __pyx_k_indexed_sentences[] = "indexed_sentences"; +static const char __pyx_k_scipy_linalg_blas[] = "scipy.linalg.blas"; +static const char __pyx_k_MAX_WORDS_IN_BATCH[] = "MAX_WORDS_IN_BATCH"; +static const char __pyx_k_cline_in_traceback[] = "cline_in_traceback"; +static const char __pyx_k_MAX_NGRAMS_IN_BATCH[] = "MAX_NGRAMS_IN_BATCH"; +static const char __pyx_k_compute_ngrams_bytes[] = "compute_ngrams_bytes"; +static const char __pyx_k_ndarray_is_not_C_contiguous[] = "ndarray is not C contiguous"; +static const char __pyx_k_gensim_models__utils_any2vec[] = "gensim.models._utils_any2vec"; +static const char __pyx_k_numpy_core_multiarray_failed_to[] = "numpy.core.multiarray failed to import"; +static const char __pyx_k_unknown_dtype_code_in_numpy_pxd[] = "unknown dtype code in numpy.pxd (%d)"; +static const char __pyx_k_Format_string_allocated_too_shor[] = "Format string allocated too short, see comment in numpy.pxd"; +static const char __pyx_k_Non_native_byte_order_not_suppor[] = "Non-native byte order not supported"; +static const char __pyx_k_Optimized_cython_functions_for_c[] = "Optimized cython functions for computing sentence embeddings"; +static const char __pyx_k_ndarray_is_not_Fortran_contiguou[] = "ndarray is not Fortran contiguous"; +static const char __pyx_k_numpy_core_umath_failed_to_impor[] = "numpy.core.umath failed to import"; +static const char __pyx_k_Format_string_allocated_too_shor_2[] = "Format string allocated too short."; +static PyObject *__pyx_n_s_FAST_VERSION; +static PyObject *__pyx_kp_u_Format_string_allocated_too_shor; +static PyObject *__pyx_kp_u_Format_string_allocated_too_shor_2; +static PyObject *__pyx_n_s_ImportError; +static PyObject *__pyx_n_s_MAX_NGRAMS_IN_BATCH; +static PyObject *__pyx_n_s_MAX_WORDS_IN_BATCH; +static PyObject *__pyx_kp_u_Non_native_byte_order_not_suppor; +static PyObject *__pyx_n_s_ONE; +static PyObject *__pyx_n_s_ONEF; +static PyObject *__pyx_n_s_RuntimeError; +static PyObject *__pyx_n_s_ValueError; +static PyObject *__pyx_n_s_ZERO; +static PyObject *__pyx_n_s_ZEROF; +static PyObject *__pyx_n_s__8; +static PyObject *__pyx_n_s_average_inner; +static PyObject *__pyx_kp_s_average_inner_pyx; +static PyObject *__pyx_n_s_bucket; +static PyObject *__pyx_n_s_cline_in_traceback; +static PyObject *__pyx_n_s_compute_ngrams_bytes; +static PyObject *__pyx_n_s_cpointer; +static PyObject *__pyx_n_s_eff_sentences; +static PyObject *__pyx_n_s_eff_words; +static PyObject *__pyx_n_s_enumerate; +static PyObject *__pyx_n_s_fblas; +static PyObject *__pyx_n_s_fill; +static PyObject *__pyx_n_s_ft; +static PyObject *__pyx_n_s_ft_hash_bytes; +static PyObject *__pyx_n_s_gensim_models__utils_any2vec; +static PyObject *__pyx_n_s_import; +static PyObject *__pyx_n_s_index; +static PyObject *__pyx_n_s_indexed_sentences; +static PyObject *__pyx_n_s_init; +static PyObject *__pyx_n_s_is_ft; +static PyObject *__pyx_n_s_main; +static PyObject *__pyx_n_s_max; +static PyObject *__pyx_n_s_max_n; +static PyObject *__pyx_n_s_memory; +static PyObject *__pyx_n_s_min_n; +static PyObject *__pyx_n_s_model; +static PyObject *__pyx_n_s_name; +static PyObject *__pyx_kp_u_ndarray_is_not_C_contiguous; +static PyObject *__pyx_kp_u_ndarray_is_not_Fortran_contiguou; +static PyObject *__pyx_n_s_np; +static PyObject *__pyx_n_s_numpy; +static PyObject *__pyx_kp_s_numpy_core_multiarray_failed_to; +static PyObject *__pyx_kp_s_numpy_core_umath_failed_to_impor; +static PyObject *__pyx_n_s_pyx_capi; +static PyObject *__pyx_n_s_range; +static PyObject *__pyx_n_s_saxpy; +static PyObject *__pyx_n_s_scipy_linalg_blas; +static PyObject *__pyx_n_s_sscal; +static PyObject *__pyx_n_s_sv; +static PyObject *__pyx_n_s_target; +static PyObject *__pyx_n_s_test; +static PyObject *__pyx_n_s_train_average_cy; +static PyObject *__pyx_kp_u_unknown_dtype_code_in_numpy_pxd; +static PyObject *__pyx_n_s_vector_size; +static PyObject *__pyx_n_s_vectors; +static PyObject *__pyx_n_s_vectors_ngrams; +static PyObject *__pyx_n_s_vectors_vocab; +static PyObject *__pyx_n_s_vocab; +static PyObject *__pyx_n_s_w2v; +static PyObject *__pyx_n_s_word_weights; +static PyObject *__pyx_n_s_workers; +static PyObject *__pyx_n_s_wv; +static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObject *__pyx_self, PyObject *__pyx_v_model, PyObject *__pyx_v_indexed_sentences, PyObject *__pyx_v_target, PyObject *__pyx_v_memory); /* proto */ +static PyObject *__pyx_pf_13average_inner_2init(CYTHON_UNUSED PyObject *__pyx_self); /* proto */ +static int __pyx_pf_5numpy_7ndarray___getbuffer__(PyArrayObject *__pyx_v_self, Py_buffer *__pyx_v_info, int __pyx_v_flags); /* proto */ +static void __pyx_pf_5numpy_7ndarray_2__releasebuffer__(PyArrayObject *__pyx_v_self, Py_buffer *__pyx_v_info); /* proto */ +static PyObject *__pyx_int_0; +static PyObject *__pyx_int_1; +static PyObject *__pyx_int_40; +static PyObject *__pyx_int_10000; +static PyObject *__pyx_tuple_; +static PyObject *__pyx_tuple__2; +static PyObject *__pyx_tuple__3; +static PyObject *__pyx_tuple__4; +static PyObject *__pyx_tuple__5; +static PyObject *__pyx_tuple__6; +static PyObject *__pyx_tuple__7; +static PyObject *__pyx_tuple__9; +static PyObject *__pyx_codeobj__10; +static PyObject *__pyx_codeobj__11; +/* Late includes */ + +/* "average_inner.pyx":36 + * DEF MAX_NGRAMS = 40 + * + * cdef init_base_s2v_config(BaseSentenceVecsConfig *c, model, target, memory): # <<<<<<<<<<<<<< + * """Load BaseAny2Vec parameters into a BaseSentenceVecsConfig struct. + * + */ + +static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, PyObject *__pyx_v_model, PyObject *__pyx_v_target, PyObject *__pyx_v_memory) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + int __pyx_t_2; + PyObject *__pyx_t_3 = NULL; + __Pyx_RefNannySetupContext("init_base_s2v_config", 0); + + /* "average_inner.pyx":52 + * + * """ + * c[0].workers = model.workers # <<<<<<<<<<<<<< + * c[0].size = model.sv.vector_size + * + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_workers); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 52, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 52, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + (__pyx_v_c[0]).workers = __pyx_t_2; + + /* "average_inner.pyx":53 + * """ + * c[0].workers = model.workers + * c[0].size = model.sv.vector_size # <<<<<<<<<<<<<< + * + * c[0].mem = (np.PyArray_DATA(memory[0])) + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_sv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 53, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vector_size); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 53, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 53, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + (__pyx_v_c[0]).size = __pyx_t_2; + + /* "average_inner.pyx":55 + * c[0].size = model.sv.vector_size + * + * c[0].mem = (np.PyArray_DATA(memory[0])) # <<<<<<<<<<<<<< + * c[0].mem2 = (np.PyArray_DATA(memory[2])) + * + */ + __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_memory, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 55, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 55, __pyx_L1_error) + (__pyx_v_c[0]).mem = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_3))); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + + /* "average_inner.pyx":56 + * + * c[0].mem = (np.PyArray_DATA(memory[0])) + * c[0].mem2 = (np.PyArray_DATA(memory[2])) # <<<<<<<<<<<<<< + * + * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors)) + */ + __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 56, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 56, __pyx_L1_error) + (__pyx_v_c[0]).mem2 = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_3))); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + + /* "average_inner.pyx":58 + * c[0].mem2 = (np.PyArray_DATA(memory[2])) + * + * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors)) # <<<<<<<<<<<<<< + * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) + * + */ + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 58, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_vectors); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 58, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 58, __pyx_L1_error) + (__pyx_v_c[0]).word_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":59 + * + * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors)) + * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) # <<<<<<<<<<<<<< + * + * c[0].sentence_vectors = (np.PyArray_DATA(target)) + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 59, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 59, __pyx_L1_error) + (__pyx_v_c[0]).word_weights = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":61 + * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) + * + * c[0].sentence_vectors = (np.PyArray_DATA(target)) # <<<<<<<<<<<<<< + * + * cdef init_ft_s2v_config(FTSentenceVecsConfig *c, model, target, memory): + */ + if (!(likely(((__pyx_v_target) == Py_None) || likely(__Pyx_TypeTest(__pyx_v_target, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 61, __pyx_L1_error) + (__pyx_v_c[0]).sentence_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_v_target))); + + /* "average_inner.pyx":36 + * DEF MAX_NGRAMS = 40 + * + * cdef init_base_s2v_config(BaseSentenceVecsConfig *c, model, target, memory): # <<<<<<<<<<<<<< + * """Load BaseAny2Vec parameters into a BaseSentenceVecsConfig struct. + * + */ + + /* function exit code */ + __pyx_r = Py_None; __Pyx_INCREF(Py_None); + goto __pyx_L0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_XDECREF(__pyx_t_3); + __Pyx_AddTraceback("average_inner.init_base_s2v_config", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "average_inner.pyx":63 + * c[0].sentence_vectors = (np.PyArray_DATA(target)) + * + * cdef init_ft_s2v_config(FTSentenceVecsConfig *c, model, target, memory): # <<<<<<<<<<<<<< + * """Load Fasttext parameters into a FTSentenceVecsConfig struct. + * + */ + +static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, PyObject *__pyx_v_model, PyObject *__pyx_v_target, PyObject *__pyx_v_memory) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + int __pyx_t_2; + PyObject *__pyx_t_3 = NULL; + PyObject *__pyx_t_4 = NULL; + PyObject *__pyx_t_5 = NULL; + __pyx_t_13average_inner_REAL_t __pyx_t_6; + __Pyx_RefNannySetupContext("init_ft_s2v_config", 0); + + /* "average_inner.pyx":80 + * """ + * + * c[0].workers = model.workers # <<<<<<<<<<<<<< + * c[0].size = model.sv.vector_size + * c[0].min_n = model.wv.min_n + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_workers); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 80, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 80, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + (__pyx_v_c[0]).workers = __pyx_t_2; + + /* "average_inner.pyx":81 + * + * c[0].workers = model.workers + * c[0].size = model.sv.vector_size # <<<<<<<<<<<<<< + * c[0].min_n = model.wv.min_n + * c[0].max_n = model.wv.max_n + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_sv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 81, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vector_size); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 81, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 81, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + (__pyx_v_c[0]).size = __pyx_t_2; + + /* "average_inner.pyx":82 + * c[0].workers = model.workers + * c[0].size = model.sv.vector_size + * c[0].min_n = model.wv.min_n # <<<<<<<<<<<<<< + * c[0].max_n = model.wv.max_n + * c[0].bucket = model.wv.bucket + */ + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 82, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_min_n); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 82, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 82, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + (__pyx_v_c[0]).min_n = __pyx_t_2; + + /* "average_inner.pyx":83 + * c[0].size = model.sv.vector_size + * c[0].min_n = model.wv.min_n + * c[0].max_n = model.wv.max_n # <<<<<<<<<<<<<< + * c[0].bucket = model.wv.bucket + * + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 83, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_max_n); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 83, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 83, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + (__pyx_v_c[0]).max_n = __pyx_t_2; + + /* "average_inner.pyx":84 + * c[0].min_n = model.wv.min_n + * c[0].max_n = model.wv.max_n + * c[0].bucket = model.wv.bucket # <<<<<<<<<<<<<< + * + * c[0].oov_weight = np.max(model.word_weights) + */ + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 84, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_bucket); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 84, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 84, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + (__pyx_v_c[0]).bucket = __pyx_t_2; + + /* "average_inner.pyx":86 + * c[0].bucket = model.wv.bucket + * + * c[0].oov_weight = np.max(model.word_weights) # <<<<<<<<<<<<<< + * + * c[0].mem = (np.PyArray_DATA(memory[0])) + */ + __Pyx_GetModuleGlobalName(__pyx_t_3, __pyx_n_s_np); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 86, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_max); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 86, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 86, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_5 = NULL; + if (CYTHON_UNPACK_METHODS && unlikely(PyMethod_Check(__pyx_t_4))) { + __pyx_t_5 = PyMethod_GET_SELF(__pyx_t_4); + if (likely(__pyx_t_5)) { + PyObject* function = PyMethod_GET_FUNCTION(__pyx_t_4); + __Pyx_INCREF(__pyx_t_5); + __Pyx_INCREF(function); + __Pyx_DECREF_SET(__pyx_t_4, function); + } + } + __pyx_t_1 = (__pyx_t_5) ? __Pyx_PyObject_Call2Args(__pyx_t_4, __pyx_t_5, __pyx_t_3) : __Pyx_PyObject_CallOneArg(__pyx_t_4, __pyx_t_3); + __Pyx_XDECREF(__pyx_t_5); __pyx_t_5 = 0; + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 86, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __pyx_PyFloat_AsFloat(__pyx_t_1); if (unlikely((__pyx_t_6 == ((npy_float32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 86, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + (__pyx_v_c[0]).oov_weight = ((__pyx_t_13average_inner_REAL_t)__pyx_t_6); + + /* "average_inner.pyx":88 + * c[0].oov_weight = np.max(model.word_weights) + * + * c[0].mem = (np.PyArray_DATA(memory[0])) # <<<<<<<<<<<<<< + * c[0].mem2 = (np.PyArray_DATA(memory[2])) + * + */ + __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 88, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 88, __pyx_L1_error) + (__pyx_v_c[0]).mem = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":89 + * + * c[0].mem = (np.PyArray_DATA(memory[0])) + * c[0].mem2 = (np.PyArray_DATA(memory[2])) # <<<<<<<<<<<<<< + * + * memory[1].fill(ZERO) # Reset the ngram storage before filling the struct + */ + __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 89, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 89, __pyx_L1_error) + (__pyx_v_c[0]).mem2 = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":91 + * c[0].mem2 = (np.PyArray_DATA(memory[2])) + * + * memory[1].fill(ZERO) # Reset the ngram storage before filling the struct # <<<<<<<<<<<<<< + * c[0].subwords_idx = (np.PyArray_DATA(memory[1])) + * + */ + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 91, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_4, __pyx_n_s_fill); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 91, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_4 = __Pyx_PyInt_From_int(__pyx_v_13average_inner_ZERO); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 91, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_5 = NULL; + if (CYTHON_UNPACK_METHODS && likely(PyMethod_Check(__pyx_t_3))) { + __pyx_t_5 = PyMethod_GET_SELF(__pyx_t_3); + if (likely(__pyx_t_5)) { + PyObject* function = PyMethod_GET_FUNCTION(__pyx_t_3); + __Pyx_INCREF(__pyx_t_5); + __Pyx_INCREF(function); + __Pyx_DECREF_SET(__pyx_t_3, function); + } + } + __pyx_t_1 = (__pyx_t_5) ? __Pyx_PyObject_Call2Args(__pyx_t_3, __pyx_t_5, __pyx_t_4) : __Pyx_PyObject_CallOneArg(__pyx_t_3, __pyx_t_4); + __Pyx_XDECREF(__pyx_t_5); __pyx_t_5 = 0; + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 91, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":92 + * + * memory[1].fill(ZERO) # Reset the ngram storage before filling the struct + * c[0].subwords_idx = (np.PyArray_DATA(memory[1])) # <<<<<<<<<<<<<< + * + * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) + */ + __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 92, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 92, __pyx_L1_error) + (__pyx_v_c[0]).subwords_idx = ((__pyx_t_13average_inner_uINT_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":94 + * c[0].subwords_idx = (np.PyArray_DATA(memory[1])) + * + * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) # <<<<<<<<<<<<<< + * c[0].ngram_vectors = (np.PyArray_DATA(model.wv.vectors_ngrams)) + * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 94, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vectors_vocab); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 94, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 94, __pyx_L1_error) + (__pyx_v_c[0]).word_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_3))); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + + /* "average_inner.pyx":95 + * + * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) + * c[0].ngram_vectors = (np.PyArray_DATA(model.wv.vectors_ngrams)) # <<<<<<<<<<<<<< + * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) + * + */ + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 95, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_vectors_ngrams); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 95, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 95, __pyx_L1_error) + (__pyx_v_c[0]).ngram_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":96 + * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) + * c[0].ngram_vectors = (np.PyArray_DATA(model.wv.vectors_ngrams)) + * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) # <<<<<<<<<<<<<< + * + * c[0].sentence_vectors = (np.PyArray_DATA(target)) + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 96, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 96, __pyx_L1_error) + (__pyx_v_c[0]).word_weights = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":98 + * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) + * + * c[0].sentence_vectors = (np.PyArray_DATA(target)) # <<<<<<<<<<<<<< + * + * cdef object populate_base_s2v_config(BaseSentenceVecsConfig *c, vocab, indexed_sentences): + */ + if (!(likely(((__pyx_v_target) == Py_None) || likely(__Pyx_TypeTest(__pyx_v_target, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 98, __pyx_L1_error) + (__pyx_v_c[0]).sentence_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_v_target))); + + /* "average_inner.pyx":63 + * c[0].sentence_vectors = (np.PyArray_DATA(target)) + * + * cdef init_ft_s2v_config(FTSentenceVecsConfig *c, model, target, memory): # <<<<<<<<<<<<<< + * """Load Fasttext parameters into a FTSentenceVecsConfig struct. + * + */ + + /* function exit code */ + __pyx_r = Py_None; __Pyx_INCREF(Py_None); + goto __pyx_L0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_XDECREF(__pyx_t_3); + __Pyx_XDECREF(__pyx_t_4); + __Pyx_XDECREF(__pyx_t_5); + __Pyx_AddTraceback("average_inner.init_ft_s2v_config", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "average_inner.pyx":100 + * c[0].sentence_vectors = (np.PyArray_DATA(target)) + * + * cdef object populate_base_s2v_config(BaseSentenceVecsConfig *c, vocab, indexed_sentences): # <<<<<<<<<<<<<< + * """Prepare C structures for BaseAny2VecModel so we can go "full C" and release the Python GIL. + * + */ + +static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, PyObject *__pyx_v_vocab, PyObject *__pyx_v_indexed_sentences) { + __pyx_t_13average_inner_uINT_t __pyx_v_eff_words; + __pyx_t_13average_inner_uINT_t __pyx_v_eff_sents; + PyObject *__pyx_v_obj = NULL; + PyObject *__pyx_v_token = NULL; + PyObject *__pyx_v_word = NULL; + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + Py_ssize_t __pyx_t_2; + PyObject *(*__pyx_t_3)(PyObject *); + PyObject *__pyx_t_4 = NULL; + int __pyx_t_5; + int __pyx_t_6; + PyObject *__pyx_t_7 = NULL; + Py_ssize_t __pyx_t_8; + PyObject *(*__pyx_t_9)(PyObject *); + PyObject *__pyx_t_10 = NULL; + __pyx_t_13average_inner_uINT_t __pyx_t_11; + __Pyx_RefNannySetupContext("populate_base_s2v_config", 0); + + /* "average_inner.pyx":124 + * """ + * + * cdef uINT_t eff_words = ZERO # Effective words encountered in a sentence # <<<<<<<<<<<<<< + * cdef uINT_t eff_sents = ZERO # Effective sentences encountered + * + */ + __pyx_v_eff_words = __pyx_v_13average_inner_ZERO; + + /* "average_inner.pyx":125 + * + * cdef uINT_t eff_words = ZERO # Effective words encountered in a sentence + * cdef uINT_t eff_sents = ZERO # Effective sentences encountered # <<<<<<<<<<<<<< + * + * c.sentence_boundary[0] = ZERO + */ + __pyx_v_eff_sents = __pyx_v_13average_inner_ZERO; + + /* "average_inner.pyx":127 + * cdef uINT_t eff_sents = ZERO # Effective sentences encountered + * + * c.sentence_boundary[0] = ZERO # <<<<<<<<<<<<<< + * + * for obj in indexed_sentences: + */ + (__pyx_v_c->sentence_boundary[0]) = __pyx_v_13average_inner_ZERO; + + /* "average_inner.pyx":129 + * c.sentence_boundary[0] = ZERO + * + * for obj in indexed_sentences: # <<<<<<<<<<<<<< + * if not obj[0]: + * continue + */ + if (likely(PyList_CheckExact(__pyx_v_indexed_sentences)) || PyTuple_CheckExact(__pyx_v_indexed_sentences)) { + __pyx_t_1 = __pyx_v_indexed_sentences; __Pyx_INCREF(__pyx_t_1); __pyx_t_2 = 0; + __pyx_t_3 = NULL; + } else { + __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 129, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext; if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 129, __pyx_L1_error) + } + for (;;) { + if (likely(!__pyx_t_3)) { + if (likely(PyList_CheckExact(__pyx_t_1))) { + if (__pyx_t_2 >= PyList_GET_SIZE(__pyx_t_1)) break; + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 129, __pyx_L1_error) + #else + __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 129, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + #endif + } else { + if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_1)) break; + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 129, __pyx_L1_error) + #else + __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 129, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + #endif + } + } else { + __pyx_t_4 = __pyx_t_3(__pyx_t_1); + if (unlikely(!__pyx_t_4)) { + PyObject* exc_type = PyErr_Occurred(); + if (exc_type) { + if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); + else __PYX_ERR(0, 129, __pyx_L1_error) + } + break; + } + __Pyx_GOTREF(__pyx_t_4); + } + __Pyx_XDECREF_SET(__pyx_v_obj, __pyx_t_4); + __pyx_t_4 = 0; + + /* "average_inner.pyx":130 + * + * for obj in indexed_sentences: + * if not obj[0]: # <<<<<<<<<<<<<< + * continue + * for token in obj[0]: + */ + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 130, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 130, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = ((!__pyx_t_5) != 0); + if (__pyx_t_6) { + + /* "average_inner.pyx":131 + * for obj in indexed_sentences: + * if not obj[0]: + * continue # <<<<<<<<<<<<<< + * for token in obj[0]: + * word = vocab[token] if token in vocab else None # Vocab obj + */ + goto __pyx_L3_continue; + + /* "average_inner.pyx":130 + * + * for obj in indexed_sentences: + * if not obj[0]: # <<<<<<<<<<<<<< + * continue + * for token in obj[0]: + */ + } + + /* "average_inner.pyx":132 + * if not obj[0]: + * continue + * for token in obj[0]: # <<<<<<<<<<<<<< + * word = vocab[token] if token in vocab else None # Vocab obj + * if word is None: + */ + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 132, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + if (likely(PyList_CheckExact(__pyx_t_4)) || PyTuple_CheckExact(__pyx_t_4)) { + __pyx_t_7 = __pyx_t_4; __Pyx_INCREF(__pyx_t_7); __pyx_t_8 = 0; + __pyx_t_9 = NULL; + } else { + __pyx_t_8 = -1; __pyx_t_7 = PyObject_GetIter(__pyx_t_4); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 132, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_7); + __pyx_t_9 = Py_TYPE(__pyx_t_7)->tp_iternext; if (unlikely(!__pyx_t_9)) __PYX_ERR(0, 132, __pyx_L1_error) + } + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + for (;;) { + if (likely(!__pyx_t_9)) { + if (likely(PyList_CheckExact(__pyx_t_7))) { + if (__pyx_t_8 >= PyList_GET_SIZE(__pyx_t_7)) break; + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_4 = PyList_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 132, __pyx_L1_error) + #else + __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 132, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + #endif + } else { + if (__pyx_t_8 >= PyTuple_GET_SIZE(__pyx_t_7)) break; + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 132, __pyx_L1_error) + #else + __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 132, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + #endif + } + } else { + __pyx_t_4 = __pyx_t_9(__pyx_t_7); + if (unlikely(!__pyx_t_4)) { + PyObject* exc_type = PyErr_Occurred(); + if (exc_type) { + if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); + else __PYX_ERR(0, 132, __pyx_L1_error) + } + break; + } + __Pyx_GOTREF(__pyx_t_4); + } + __Pyx_XDECREF_SET(__pyx_v_token, __pyx_t_4); + __pyx_t_4 = 0; + + /* "average_inner.pyx":133 + * continue + * for token in obj[0]: + * word = vocab[token] if token in vocab else None # Vocab obj # <<<<<<<<<<<<<< + * if word is None: + * continue + */ + __pyx_t_6 = (__Pyx_PySequence_ContainsTF(__pyx_v_token, __pyx_v_vocab, Py_EQ)); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(0, 133, __pyx_L1_error) + if ((__pyx_t_6 != 0)) { + __pyx_t_10 = __Pyx_PyObject_GetItem(__pyx_v_vocab, __pyx_v_token); if (unlikely(!__pyx_t_10)) __PYX_ERR(0, 133, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_10); + __pyx_t_4 = __pyx_t_10; + __pyx_t_10 = 0; + } else { + __Pyx_INCREF(Py_None); + __pyx_t_4 = Py_None; + } + __Pyx_XDECREF_SET(__pyx_v_word, __pyx_t_4); + __pyx_t_4 = 0; + + /* "average_inner.pyx":134 + * for token in obj[0]: + * word = vocab[token] if token in vocab else None # Vocab obj + * if word is None: # <<<<<<<<<<<<<< + * continue + * c.word_indices[eff_words] = word.index + */ + __pyx_t_6 = (__pyx_v_word == Py_None); + __pyx_t_5 = (__pyx_t_6 != 0); + if (__pyx_t_5) { + + /* "average_inner.pyx":135 + * word = vocab[token] if token in vocab else None # Vocab obj + * if word is None: + * continue # <<<<<<<<<<<<<< + * c.word_indices[eff_words] = word.index + * c.sent_adresses[eff_words] = obj[1] + */ + goto __pyx_L6_continue; + + /* "average_inner.pyx":134 + * for token in obj[0]: + * word = vocab[token] if token in vocab else None # Vocab obj + * if word is None: # <<<<<<<<<<<<<< + * continue + * c.word_indices[eff_words] = word.index + */ + } + + /* "average_inner.pyx":136 + * if word is None: + * continue + * c.word_indices[eff_words] = word.index # <<<<<<<<<<<<<< + * c.sent_adresses[eff_words] = obj[1] + * + */ + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_v_word, __pyx_n_s_index); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 136, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_11 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_11 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 136, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + (__pyx_v_c->word_indices[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_11); + + /* "average_inner.pyx":137 + * continue + * c.word_indices[eff_words] = word.index + * c.sent_adresses[eff_words] = obj[1] # <<<<<<<<<<<<<< + * + * eff_words += ONE + */ + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 137, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_11 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_11 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 137, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + (__pyx_v_c->sent_adresses[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_11); + + /* "average_inner.pyx":139 + * c.sent_adresses[eff_words] = obj[1] + * + * eff_words += ONE # <<<<<<<<<<<<<< + * if eff_words == MAX_WORDS: + * break + */ + __pyx_v_eff_words = (__pyx_v_eff_words + __pyx_v_13average_inner_ONE); + + /* "average_inner.pyx":140 + * + * eff_words += ONE + * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< + * break + * eff_sents += 1 + */ + __pyx_t_5 = ((__pyx_v_eff_words == 0x2710) != 0); + if (__pyx_t_5) { + + /* "average_inner.pyx":141 + * eff_words += ONE + * if eff_words == MAX_WORDS: + * break # <<<<<<<<<<<<<< + * eff_sents += 1 + * c.sentence_boundary[eff_sents] = eff_words + */ + goto __pyx_L7_break; + + /* "average_inner.pyx":140 + * + * eff_words += ONE + * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< + * break + * eff_sents += 1 + */ + } + + /* "average_inner.pyx":132 + * if not obj[0]: + * continue + * for token in obj[0]: # <<<<<<<<<<<<<< + * word = vocab[token] if token in vocab else None # Vocab obj + * if word is None: + */ + __pyx_L6_continue:; + } + __pyx_L7_break:; + __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; + + /* "average_inner.pyx":142 + * if eff_words == MAX_WORDS: + * break + * eff_sents += 1 # <<<<<<<<<<<<<< + * c.sentence_boundary[eff_sents] = eff_words + * + */ + __pyx_v_eff_sents = (__pyx_v_eff_sents + 1); + + /* "average_inner.pyx":143 + * break + * eff_sents += 1 + * c.sentence_boundary[eff_sents] = eff_words # <<<<<<<<<<<<<< + * + * if eff_words == MAX_WORDS: + */ + (__pyx_v_c->sentence_boundary[__pyx_v_eff_sents]) = __pyx_v_eff_words; + + /* "average_inner.pyx":145 + * c.sentence_boundary[eff_sents] = eff_words + * + * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< + * break + * + */ + __pyx_t_5 = ((__pyx_v_eff_words == 0x2710) != 0); + if (__pyx_t_5) { + + /* "average_inner.pyx":146 + * + * if eff_words == MAX_WORDS: + * break # <<<<<<<<<<<<<< + * + * return eff_sents, eff_words + */ + goto __pyx_L4_break; + + /* "average_inner.pyx":145 + * c.sentence_boundary[eff_sents] = eff_words + * + * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< + * break + * + */ + } + + /* "average_inner.pyx":129 + * c.sentence_boundary[0] = ZERO + * + * for obj in indexed_sentences: # <<<<<<<<<<<<<< + * if not obj[0]: + * continue + */ + __pyx_L3_continue:; + } + __pyx_L4_break:; + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":148 + * break + * + * return eff_sents, eff_words # <<<<<<<<<<<<<< + * + * cdef object populate_ft_s2v_config(FTSentenceVecsConfig *c, vocab, indexed_sentences): + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sents); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 148, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_7 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 148, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_7); + __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 148, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __Pyx_GIVEREF(__pyx_t_1); + PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_1); + __Pyx_GIVEREF(__pyx_t_7); + PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_t_7); + __pyx_t_1 = 0; + __pyx_t_7 = 0; + __pyx_r = __pyx_t_4; + __pyx_t_4 = 0; + goto __pyx_L0; + + /* "average_inner.pyx":100 + * c[0].sentence_vectors = (np.PyArray_DATA(target)) + * + * cdef object populate_base_s2v_config(BaseSentenceVecsConfig *c, vocab, indexed_sentences): # <<<<<<<<<<<<<< + * """Prepare C structures for BaseAny2VecModel so we can go "full C" and release the Python GIL. + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_XDECREF(__pyx_t_4); + __Pyx_XDECREF(__pyx_t_7); + __Pyx_XDECREF(__pyx_t_10); + __Pyx_AddTraceback("average_inner.populate_base_s2v_config", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XDECREF(__pyx_v_obj); + __Pyx_XDECREF(__pyx_v_token); + __Pyx_XDECREF(__pyx_v_word); + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "average_inner.pyx":150 + * return eff_sents, eff_words + * + * cdef object populate_ft_s2v_config(FTSentenceVecsConfig *c, vocab, indexed_sentences): # <<<<<<<<<<<<<< + * """Prepare C structures for FastText so we can go "full C" and release the Python GIL. + * + */ + +static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, PyObject *__pyx_v_vocab, PyObject *__pyx_v_indexed_sentences) { + __pyx_t_13average_inner_uINT_t __pyx_v_eff_words; + __pyx_t_13average_inner_uINT_t __pyx_v_eff_sents; + PyObject *__pyx_v_obj = NULL; + PyObject *__pyx_v_token = NULL; + PyObject *__pyx_v_word = NULL; + PyObject *__pyx_v_encoded_ngrams = NULL; + PyObject *__pyx_v_hashes = NULL; + PyObject *__pyx_v_i = NULL; + PyObject *__pyx_v_h = NULL; + PyObject *__pyx_v_n = NULL; + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + Py_ssize_t __pyx_t_2; + PyObject *(*__pyx_t_3)(PyObject *); + PyObject *__pyx_t_4 = NULL; + int __pyx_t_5; + int __pyx_t_6; + PyObject *__pyx_t_7 = NULL; + Py_ssize_t __pyx_t_8; + PyObject *(*__pyx_t_9)(PyObject *); + __pyx_t_13average_inner_uINT_t __pyx_t_10; + PyObject *__pyx_t_11 = NULL; + PyObject *__pyx_t_12 = NULL; + PyObject *__pyx_t_13 = NULL; + PyObject *__pyx_t_14 = NULL; + int __pyx_t_15; + PyObject *__pyx_t_16 = NULL; + Py_ssize_t __pyx_t_17; + PyObject *(*__pyx_t_18)(PyObject *); + long __pyx_t_19; + Py_ssize_t __pyx_t_20; + __Pyx_RefNannySetupContext("populate_ft_s2v_config", 0); + + /* "average_inner.pyx":174 + * """ + * + * cdef uINT_t eff_words = ZERO # Effective words encountered in a sentence # <<<<<<<<<<<<<< + * cdef uINT_t eff_sents = ZERO # Effective sentences encountered + * + */ + __pyx_v_eff_words = __pyx_v_13average_inner_ZERO; + + /* "average_inner.pyx":175 + * + * cdef uINT_t eff_words = ZERO # Effective words encountered in a sentence + * cdef uINT_t eff_sents = ZERO # Effective sentences encountered # <<<<<<<<<<<<<< + * + * c.sentence_boundary[0] = ZERO + */ + __pyx_v_eff_sents = __pyx_v_13average_inner_ZERO; + + /* "average_inner.pyx":177 + * cdef uINT_t eff_sents = ZERO # Effective sentences encountered + * + * c.sentence_boundary[0] = ZERO # <<<<<<<<<<<<<< + * + * for obj in indexed_sentences: + */ + (__pyx_v_c->sentence_boundary[0]) = __pyx_v_13average_inner_ZERO; + + /* "average_inner.pyx":179 + * c.sentence_boundary[0] = ZERO + * + * for obj in indexed_sentences: # <<<<<<<<<<<<<< + * if not obj[0]: + * continue + */ + if (likely(PyList_CheckExact(__pyx_v_indexed_sentences)) || PyTuple_CheckExact(__pyx_v_indexed_sentences)) { + __pyx_t_1 = __pyx_v_indexed_sentences; __Pyx_INCREF(__pyx_t_1); __pyx_t_2 = 0; + __pyx_t_3 = NULL; + } else { + __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 179, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext; if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 179, __pyx_L1_error) + } + for (;;) { + if (likely(!__pyx_t_3)) { + if (likely(PyList_CheckExact(__pyx_t_1))) { + if (__pyx_t_2 >= PyList_GET_SIZE(__pyx_t_1)) break; + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 179, __pyx_L1_error) + #else + __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 179, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + #endif + } else { + if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_1)) break; + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 179, __pyx_L1_error) + #else + __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 179, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + #endif + } + } else { + __pyx_t_4 = __pyx_t_3(__pyx_t_1); + if (unlikely(!__pyx_t_4)) { + PyObject* exc_type = PyErr_Occurred(); + if (exc_type) { + if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); + else __PYX_ERR(0, 179, __pyx_L1_error) + } + break; + } + __Pyx_GOTREF(__pyx_t_4); + } + __Pyx_XDECREF_SET(__pyx_v_obj, __pyx_t_4); + __pyx_t_4 = 0; + + /* "average_inner.pyx":180 + * + * for obj in indexed_sentences: + * if not obj[0]: # <<<<<<<<<<<<<< + * continue + * for token in obj[0]: + */ + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 180, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 180, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = ((!__pyx_t_5) != 0); + if (__pyx_t_6) { + + /* "average_inner.pyx":181 + * for obj in indexed_sentences: + * if not obj[0]: + * continue # <<<<<<<<<<<<<< + * for token in obj[0]: + * c.sent_adresses[eff_words] = obj[1] + */ + goto __pyx_L3_continue; + + /* "average_inner.pyx":180 + * + * for obj in indexed_sentences: + * if not obj[0]: # <<<<<<<<<<<<<< + * continue + * for token in obj[0]: + */ + } + + /* "average_inner.pyx":182 + * if not obj[0]: + * continue + * for token in obj[0]: # <<<<<<<<<<<<<< + * c.sent_adresses[eff_words] = obj[1] + * if token in vocab: + */ + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 182, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + if (likely(PyList_CheckExact(__pyx_t_4)) || PyTuple_CheckExact(__pyx_t_4)) { + __pyx_t_7 = __pyx_t_4; __Pyx_INCREF(__pyx_t_7); __pyx_t_8 = 0; + __pyx_t_9 = NULL; + } else { + __pyx_t_8 = -1; __pyx_t_7 = PyObject_GetIter(__pyx_t_4); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 182, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_7); + __pyx_t_9 = Py_TYPE(__pyx_t_7)->tp_iternext; if (unlikely(!__pyx_t_9)) __PYX_ERR(0, 182, __pyx_L1_error) + } + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + for (;;) { + if (likely(!__pyx_t_9)) { + if (likely(PyList_CheckExact(__pyx_t_7))) { + if (__pyx_t_8 >= PyList_GET_SIZE(__pyx_t_7)) break; + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_4 = PyList_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 182, __pyx_L1_error) + #else + __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 182, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + #endif + } else { + if (__pyx_t_8 >= PyTuple_GET_SIZE(__pyx_t_7)) break; + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 182, __pyx_L1_error) + #else + __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 182, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + #endif + } + } else { + __pyx_t_4 = __pyx_t_9(__pyx_t_7); + if (unlikely(!__pyx_t_4)) { + PyObject* exc_type = PyErr_Occurred(); + if (exc_type) { + if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); + else __PYX_ERR(0, 182, __pyx_L1_error) + } + break; + } + __Pyx_GOTREF(__pyx_t_4); + } + __Pyx_XDECREF_SET(__pyx_v_token, __pyx_t_4); + __pyx_t_4 = 0; + + /* "average_inner.pyx":183 + * continue + * for token in obj[0]: + * c.sent_adresses[eff_words] = obj[1] # <<<<<<<<<<<<<< + * if token in vocab: + * # In Vocabulary + */ + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 183, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 183, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + (__pyx_v_c->sent_adresses[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_10); + + /* "average_inner.pyx":184 + * for token in obj[0]: + * c.sent_adresses[eff_words] = obj[1] + * if token in vocab: # <<<<<<<<<<<<<< + * # In Vocabulary + * word = vocab[token] + */ + __pyx_t_6 = (__Pyx_PySequence_ContainsTF(__pyx_v_token, __pyx_v_vocab, Py_EQ)); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(0, 184, __pyx_L1_error) + __pyx_t_5 = (__pyx_t_6 != 0); + if (__pyx_t_5) { + + /* "average_inner.pyx":186 + * if token in vocab: + * # In Vocabulary + * word = vocab[token] # <<<<<<<<<<<<<< + * c.word_indices[eff_words] = word.index + * c.subwords_idx_len[eff_words] = ZERO + */ + __pyx_t_4 = __Pyx_PyObject_GetItem(__pyx_v_vocab, __pyx_v_token); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 186, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __Pyx_XDECREF_SET(__pyx_v_word, __pyx_t_4); + __pyx_t_4 = 0; + + /* "average_inner.pyx":187 + * # In Vocabulary + * word = vocab[token] + * c.word_indices[eff_words] = word.index # <<<<<<<<<<<<<< + * c.subwords_idx_len[eff_words] = ZERO + * else: + */ + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_v_word, __pyx_n_s_index); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 187, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 187, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + (__pyx_v_c->word_indices[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_10); + + /* "average_inner.pyx":188 + * word = vocab[token] + * c.word_indices[eff_words] = word.index + * c.subwords_idx_len[eff_words] = ZERO # <<<<<<<<<<<<<< + * else: + * # OOV words --> write ngram indices to memory + */ + (__pyx_v_c->subwords_idx_len[__pyx_v_eff_words]) = __pyx_v_13average_inner_ZERO; + + /* "average_inner.pyx":184 + * for token in obj[0]: + * c.sent_adresses[eff_words] = obj[1] + * if token in vocab: # <<<<<<<<<<<<<< + * # In Vocabulary + * word = vocab[token] + */ + goto __pyx_L8; + } + + /* "average_inner.pyx":191 + * else: + * # OOV words --> write ngram indices to memory + * c.word_indices[eff_words] = ZERO # <<<<<<<<<<<<<< + * + * encoded_ngrams = compute_ngrams_bytes(token, c.min_n, c.max_n) + */ + /*else*/ { + (__pyx_v_c->word_indices[__pyx_v_eff_words]) = __pyx_v_13average_inner_ZERO; + + /* "average_inner.pyx":193 + * c.word_indices[eff_words] = ZERO + * + * encoded_ngrams = compute_ngrams_bytes(token, c.min_n, c.max_n) # <<<<<<<<<<<<<< + * hashes = [ft_hash_bytes(n) % c.bucket for n in encoded_ngrams] + * + */ + __Pyx_GetModuleGlobalName(__pyx_t_11, __pyx_n_s_compute_ngrams_bytes); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 193, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_11); + __pyx_t_12 = __Pyx_PyInt_From_int(__pyx_v_c->min_n); if (unlikely(!__pyx_t_12)) __PYX_ERR(0, 193, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_12); + __pyx_t_13 = __Pyx_PyInt_From_int(__pyx_v_c->max_n); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 193, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_13); + __pyx_t_14 = NULL; + __pyx_t_15 = 0; + if (CYTHON_UNPACK_METHODS && unlikely(PyMethod_Check(__pyx_t_11))) { + __pyx_t_14 = PyMethod_GET_SELF(__pyx_t_11); + if (likely(__pyx_t_14)) { + PyObject* function = PyMethod_GET_FUNCTION(__pyx_t_11); + __Pyx_INCREF(__pyx_t_14); + __Pyx_INCREF(function); + __Pyx_DECREF_SET(__pyx_t_11, function); + __pyx_t_15 = 1; + } + } + #if CYTHON_FAST_PYCALL + if (PyFunction_Check(__pyx_t_11)) { + PyObject *__pyx_temp[4] = {__pyx_t_14, __pyx_v_token, __pyx_t_12, __pyx_t_13}; + __pyx_t_4 = __Pyx_PyFunction_FastCall(__pyx_t_11, __pyx_temp+1-__pyx_t_15, 3+__pyx_t_15); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 193, __pyx_L1_error) + __Pyx_XDECREF(__pyx_t_14); __pyx_t_14 = 0; + __Pyx_GOTREF(__pyx_t_4); + __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0; + __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0; + } else + #endif + #if CYTHON_FAST_PYCCALL + if (__Pyx_PyFastCFunction_Check(__pyx_t_11)) { + PyObject *__pyx_temp[4] = {__pyx_t_14, __pyx_v_token, __pyx_t_12, __pyx_t_13}; + __pyx_t_4 = __Pyx_PyCFunction_FastCall(__pyx_t_11, __pyx_temp+1-__pyx_t_15, 3+__pyx_t_15); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 193, __pyx_L1_error) + __Pyx_XDECREF(__pyx_t_14); __pyx_t_14 = 0; + __Pyx_GOTREF(__pyx_t_4); + __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0; + __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0; + } else + #endif + { + __pyx_t_16 = PyTuple_New(3+__pyx_t_15); if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 193, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_16); + if (__pyx_t_14) { + __Pyx_GIVEREF(__pyx_t_14); PyTuple_SET_ITEM(__pyx_t_16, 0, __pyx_t_14); __pyx_t_14 = NULL; + } + __Pyx_INCREF(__pyx_v_token); + __Pyx_GIVEREF(__pyx_v_token); + PyTuple_SET_ITEM(__pyx_t_16, 0+__pyx_t_15, __pyx_v_token); + __Pyx_GIVEREF(__pyx_t_12); + PyTuple_SET_ITEM(__pyx_t_16, 1+__pyx_t_15, __pyx_t_12); + __Pyx_GIVEREF(__pyx_t_13); + PyTuple_SET_ITEM(__pyx_t_16, 2+__pyx_t_15, __pyx_t_13); + __pyx_t_12 = 0; + __pyx_t_13 = 0; + __pyx_t_4 = __Pyx_PyObject_Call(__pyx_t_11, __pyx_t_16, NULL); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 193, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __Pyx_DECREF(__pyx_t_16); __pyx_t_16 = 0; + } + __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 0; + __Pyx_XDECREF_SET(__pyx_v_encoded_ngrams, __pyx_t_4); + __pyx_t_4 = 0; + + /* "average_inner.pyx":194 + * + * encoded_ngrams = compute_ngrams_bytes(token, c.min_n, c.max_n) + * hashes = [ft_hash_bytes(n) % c.bucket for n in encoded_ngrams] # <<<<<<<<<<<<<< + * + * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) + */ + __pyx_t_4 = PyList_New(0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 194, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + if (likely(PyList_CheckExact(__pyx_v_encoded_ngrams)) || PyTuple_CheckExact(__pyx_v_encoded_ngrams)) { + __pyx_t_11 = __pyx_v_encoded_ngrams; __Pyx_INCREF(__pyx_t_11); __pyx_t_17 = 0; + __pyx_t_18 = NULL; + } else { + __pyx_t_17 = -1; __pyx_t_11 = PyObject_GetIter(__pyx_v_encoded_ngrams); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 194, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_11); + __pyx_t_18 = Py_TYPE(__pyx_t_11)->tp_iternext; if (unlikely(!__pyx_t_18)) __PYX_ERR(0, 194, __pyx_L1_error) + } + for (;;) { + if (likely(!__pyx_t_18)) { + if (likely(PyList_CheckExact(__pyx_t_11))) { + if (__pyx_t_17 >= PyList_GET_SIZE(__pyx_t_11)) break; + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_16 = PyList_GET_ITEM(__pyx_t_11, __pyx_t_17); __Pyx_INCREF(__pyx_t_16); __pyx_t_17++; if (unlikely(0 < 0)) __PYX_ERR(0, 194, __pyx_L1_error) + #else + __pyx_t_16 = PySequence_ITEM(__pyx_t_11, __pyx_t_17); __pyx_t_17++; if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 194, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_16); + #endif + } else { + if (__pyx_t_17 >= PyTuple_GET_SIZE(__pyx_t_11)) break; + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_16 = PyTuple_GET_ITEM(__pyx_t_11, __pyx_t_17); __Pyx_INCREF(__pyx_t_16); __pyx_t_17++; if (unlikely(0 < 0)) __PYX_ERR(0, 194, __pyx_L1_error) + #else + __pyx_t_16 = PySequence_ITEM(__pyx_t_11, __pyx_t_17); __pyx_t_17++; if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 194, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_16); + #endif + } + } else { + __pyx_t_16 = __pyx_t_18(__pyx_t_11); + if (unlikely(!__pyx_t_16)) { + PyObject* exc_type = PyErr_Occurred(); + if (exc_type) { + if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); + else __PYX_ERR(0, 194, __pyx_L1_error) + } + break; + } + __Pyx_GOTREF(__pyx_t_16); + } + __Pyx_XDECREF_SET(__pyx_v_n, __pyx_t_16); + __pyx_t_16 = 0; + __Pyx_GetModuleGlobalName(__pyx_t_13, __pyx_n_s_ft_hash_bytes); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 194, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_13); + __pyx_t_12 = NULL; + if (CYTHON_UNPACK_METHODS && unlikely(PyMethod_Check(__pyx_t_13))) { + __pyx_t_12 = PyMethod_GET_SELF(__pyx_t_13); + if (likely(__pyx_t_12)) { + PyObject* function = PyMethod_GET_FUNCTION(__pyx_t_13); + __Pyx_INCREF(__pyx_t_12); + __Pyx_INCREF(function); + __Pyx_DECREF_SET(__pyx_t_13, function); + } + } + __pyx_t_16 = (__pyx_t_12) ? __Pyx_PyObject_Call2Args(__pyx_t_13, __pyx_t_12, __pyx_v_n) : __Pyx_PyObject_CallOneArg(__pyx_t_13, __pyx_v_n); + __Pyx_XDECREF(__pyx_t_12); __pyx_t_12 = 0; + if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 194, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_16); + __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0; + __pyx_t_13 = __Pyx_PyInt_From_int(__pyx_v_c->bucket); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 194, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_13); + __pyx_t_12 = PyNumber_Remainder(__pyx_t_16, __pyx_t_13); if (unlikely(!__pyx_t_12)) __PYX_ERR(0, 194, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_12); + __Pyx_DECREF(__pyx_t_16); __pyx_t_16 = 0; + __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0; + if (unlikely(__Pyx_ListComp_Append(__pyx_t_4, (PyObject*)__pyx_t_12))) __PYX_ERR(0, 194, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0; + } + __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 0; + __Pyx_XDECREF_SET(__pyx_v_hashes, ((PyObject*)__pyx_t_4)); + __pyx_t_4 = 0; + + /* "average_inner.pyx":196 + * hashes = [ft_hash_bytes(n) % c.bucket for n in encoded_ngrams] + * + * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) # <<<<<<<<<<<<<< + * for i, h in enumerate(hashes[:MAX_NGRAMS]): + * c.subwords_idx[(eff_words * MAX_NGRAMS) + i] = h + */ + __pyx_t_19 = 40; + __pyx_t_17 = PyObject_Length(__pyx_v_encoded_ngrams); if (unlikely(__pyx_t_17 == ((Py_ssize_t)-1))) __PYX_ERR(0, 196, __pyx_L1_error) + if (((__pyx_t_19 < __pyx_t_17) != 0)) { + __pyx_t_20 = __pyx_t_19; + } else { + __pyx_t_20 = __pyx_t_17; + } + (__pyx_v_c->subwords_idx_len[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_20); + + /* "average_inner.pyx":197 + * + * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) + * for i, h in enumerate(hashes[:MAX_NGRAMS]): # <<<<<<<<<<<<<< + * c.subwords_idx[(eff_words * MAX_NGRAMS) + i] = h + * + */ + __Pyx_INCREF(__pyx_int_0); + __pyx_t_4 = __pyx_int_0; + __pyx_t_11 = __Pyx_PyList_GetSlice(__pyx_v_hashes, 0, 40); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 197, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_11); + __pyx_t_12 = __pyx_t_11; __Pyx_INCREF(__pyx_t_12); __pyx_t_20 = 0; + __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 0; + for (;;) { + if (__pyx_t_20 >= PyList_GET_SIZE(__pyx_t_12)) break; + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_11 = PyList_GET_ITEM(__pyx_t_12, __pyx_t_20); __Pyx_INCREF(__pyx_t_11); __pyx_t_20++; if (unlikely(0 < 0)) __PYX_ERR(0, 197, __pyx_L1_error) + #else + __pyx_t_11 = PySequence_ITEM(__pyx_t_12, __pyx_t_20); __pyx_t_20++; if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 197, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_11); + #endif + __Pyx_XDECREF_SET(__pyx_v_h, __pyx_t_11); + __pyx_t_11 = 0; + __Pyx_INCREF(__pyx_t_4); + __Pyx_XDECREF_SET(__pyx_v_i, __pyx_t_4); + __pyx_t_11 = __Pyx_PyInt_AddObjC(__pyx_t_4, __pyx_int_1, 1, 0, 0); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 197, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_11); + __Pyx_DECREF(__pyx_t_4); + __pyx_t_4 = __pyx_t_11; + __pyx_t_11 = 0; + + /* "average_inner.pyx":198 + * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) + * for i, h in enumerate(hashes[:MAX_NGRAMS]): + * c.subwords_idx[(eff_words * MAX_NGRAMS) + i] = h # <<<<<<<<<<<<<< + * + * eff_words += ONE + */ + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_v_h); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 198, __pyx_L1_error) + __pyx_t_11 = __Pyx_PyInt_From_long((__pyx_v_eff_words * 40)); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 198, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_11); + __pyx_t_13 = PyNumber_Add(__pyx_t_11, __pyx_v_i); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 198, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_13); + __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 0; + __pyx_t_17 = __Pyx_PyIndex_AsSsize_t(__pyx_t_13); if (unlikely((__pyx_t_17 == (Py_ssize_t)-1) && PyErr_Occurred())) __PYX_ERR(0, 198, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0; + (__pyx_v_c->subwords_idx[__pyx_t_17]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_10); + + /* "average_inner.pyx":197 + * + * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) + * for i, h in enumerate(hashes[:MAX_NGRAMS]): # <<<<<<<<<<<<<< + * c.subwords_idx[(eff_words * MAX_NGRAMS) + i] = h + * + */ + } + __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0; + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + } + __pyx_L8:; + + /* "average_inner.pyx":200 + * c.subwords_idx[(eff_words * MAX_NGRAMS) + i] = h + * + * eff_words += ONE # <<<<<<<<<<<<<< + * + * if eff_words == MAX_WORDS: + */ + __pyx_v_eff_words = (__pyx_v_eff_words + __pyx_v_13average_inner_ONE); + + /* "average_inner.pyx":202 + * eff_words += ONE + * + * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< + * break + * + */ + __pyx_t_5 = ((__pyx_v_eff_words == 0x2710) != 0); + if (__pyx_t_5) { + + /* "average_inner.pyx":203 + * + * if eff_words == MAX_WORDS: + * break # <<<<<<<<<<<<<< + * + * eff_sents += 1 + */ + goto __pyx_L7_break; + + /* "average_inner.pyx":202 + * eff_words += ONE + * + * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< + * break + * + */ + } + + /* "average_inner.pyx":182 + * if not obj[0]: + * continue + * for token in obj[0]: # <<<<<<<<<<<<<< + * c.sent_adresses[eff_words] = obj[1] + * if token in vocab: + */ + } + __pyx_L7_break:; + __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; + + /* "average_inner.pyx":205 + * break + * + * eff_sents += 1 # <<<<<<<<<<<<<< + * c.sentence_boundary[eff_sents] = eff_words + * + */ + __pyx_v_eff_sents = (__pyx_v_eff_sents + 1); + + /* "average_inner.pyx":206 + * + * eff_sents += 1 + * c.sentence_boundary[eff_sents] = eff_words # <<<<<<<<<<<<<< + * + * if eff_words == MAX_WORDS: + */ + (__pyx_v_c->sentence_boundary[__pyx_v_eff_sents]) = __pyx_v_eff_words; + + /* "average_inner.pyx":208 + * c.sentence_boundary[eff_sents] = eff_words + * + * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< + * break + * + */ + __pyx_t_5 = ((__pyx_v_eff_words == 0x2710) != 0); + if (__pyx_t_5) { + + /* "average_inner.pyx":209 + * + * if eff_words == MAX_WORDS: + * break # <<<<<<<<<<<<<< + * + * return eff_sents, eff_words + */ + goto __pyx_L4_break; + + /* "average_inner.pyx":208 + * c.sentence_boundary[eff_sents] = eff_words + * + * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< + * break + * + */ + } + + /* "average_inner.pyx":179 + * c.sentence_boundary[0] = ZERO + * + * for obj in indexed_sentences: # <<<<<<<<<<<<<< + * if not obj[0]: + * continue + */ + __pyx_L3_continue:; + } + __pyx_L4_break:; + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":211 + * break + * + * return eff_sents, eff_words # <<<<<<<<<<<<<< + * + * cdef void compute_base_sentence_averages(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sents); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 211, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_7 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 211, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_7); + __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 211, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __Pyx_GIVEREF(__pyx_t_1); + PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_1); + __Pyx_GIVEREF(__pyx_t_7); + PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_t_7); + __pyx_t_1 = 0; + __pyx_t_7 = 0; + __pyx_r = __pyx_t_4; + __pyx_t_4 = 0; + goto __pyx_L0; + + /* "average_inner.pyx":150 + * return eff_sents, eff_words + * + * cdef object populate_ft_s2v_config(FTSentenceVecsConfig *c, vocab, indexed_sentences): # <<<<<<<<<<<<<< + * """Prepare C structures for FastText so we can go "full C" and release the Python GIL. + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_XDECREF(__pyx_t_4); + __Pyx_XDECREF(__pyx_t_7); + __Pyx_XDECREF(__pyx_t_11); + __Pyx_XDECREF(__pyx_t_12); + __Pyx_XDECREF(__pyx_t_13); + __Pyx_XDECREF(__pyx_t_14); + __Pyx_XDECREF(__pyx_t_16); + __Pyx_AddTraceback("average_inner.populate_ft_s2v_config", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XDECREF(__pyx_v_obj); + __Pyx_XDECREF(__pyx_v_token); + __Pyx_XDECREF(__pyx_v_word); + __Pyx_XDECREF(__pyx_v_encoded_ngrams); + __Pyx_XDECREF(__pyx_v_hashes); + __Pyx_XDECREF(__pyx_v_i); + __Pyx_XDECREF(__pyx_v_h); + __Pyx_XDECREF(__pyx_v_n); + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "average_inner.pyx":213 + * return eff_sents, eff_words + * + * cdef void compute_base_sentence_averages(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< + * """Perform optimized sentence-level averaging for BaseAny2Vec model. + * + */ + +static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { + int __pyx_v_size; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_end; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; + __pyx_t_13average_inner_uINT_t __pyx_v_i; + __pyx_t_13average_inner_uINT_t __pyx_v_word_idx; + __pyx_t_13average_inner_uINT_t __pyx_v_word_row; + __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; + __pyx_t_13average_inner_REAL_t __pyx_v_inv_count; + int __pyx_t_1; + __pyx_t_13average_inner_uINT_t __pyx_t_2; + __pyx_t_13average_inner_uINT_t __pyx_t_3; + __pyx_t_13average_inner_uINT_t __pyx_t_4; + __pyx_t_13average_inner_uINT_t __pyx_t_5; + __pyx_t_13average_inner_uINT_t __pyx_t_6; + __pyx_t_13average_inner_uINT_t __pyx_t_7; + int __pyx_t_8; + + /* "average_inner.pyx":229 + * """ + * cdef: + * int size = c.size # <<<<<<<<<<<<<< + * + * uINT_t sent_idx, sent_start, sent_end, sent_row + */ + __pyx_t_1 = __pyx_v_c->size; + __pyx_v_size = __pyx_t_1; + + /* "average_inner.pyx":237 + * REAL_t sent_len, inv_count + * + * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * + */ + __pyx_t_2 = __pyx_v_num_sentences; + __pyx_t_3 = __pyx_t_2; + for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) { + __pyx_v_sent_idx = __pyx_t_4; + + /* "average_inner.pyx":238 + * + * for sent_idx in range(num_sentences): + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< + * + * sent_start = c.sentence_boundary[sent_idx] + */ + (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); + + /* "average_inner.pyx":240 + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * + * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< + * sent_end = c.sentence_boundary[sent_idx + 1] + * sent_len = ZEROF + */ + __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); + + /* "average_inner.pyx":241 + * + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< + * sent_len = ZEROF + * + */ + __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); + + /* "average_inner.pyx":242 + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] + * sent_len = ZEROF # <<<<<<<<<<<<<< + * + * for i in range(sent_start, sent_end): + */ + __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; + + /* "average_inner.pyx":244 + * sent_len = ZEROF + * + * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< + * sent_len += ONEF + * sent_row = c.sent_adresses[i] * size + */ + __pyx_t_5 = __pyx_v_sent_end; + __pyx_t_6 = __pyx_t_5; + for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { + __pyx_v_i = __pyx_t_7; + + /* "average_inner.pyx":245 + * + * for i in range(sent_start, sent_end): + * sent_len += ONEF # <<<<<<<<<<<<<< + * sent_row = c.sent_adresses[i] * size + * word_row = c.word_indices[i] * size + */ + __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); + + /* "average_inner.pyx":246 + * for i in range(sent_start, sent_end): + * sent_len += ONEF + * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< + * word_row = c.word_indices[i] * size + * word_idx = c.word_indices[i] + */ + __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); + + /* "average_inner.pyx":247 + * sent_len += ONEF + * sent_row = c.sent_adresses[i] * size + * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< + * word_idx = c.word_indices[i] + * + */ + __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); + + /* "average_inner.pyx":248 + * sent_row = c.sent_adresses[i] * size + * word_row = c.word_indices[i] * size + * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< + * + * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + */ + __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); + + /* "average_inner.pyx":250 + * word_idx = c.word_indices[i] + * + * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) # <<<<<<<<<<<<<< + * + * if sent_len > ZEROF: + */ + __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); + } + + /* "average_inner.pyx":252 + * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + * + * if sent_len > ZEROF: # <<<<<<<<<<<<<< + * inv_count = ONEF / sent_len + * # If we perform the a*x on memory, the computation is compatible with many-to-one mappings + */ + __pyx_t_8 = ((__pyx_v_sent_len > __pyx_v_13average_inner_ZEROF) != 0); + if (__pyx_t_8) { + + /* "average_inner.pyx":253 + * + * if sent_len > ZEROF: + * inv_count = ONEF / sent_len # <<<<<<<<<<<<<< + * # If we perform the a*x on memory, the computation is compatible with many-to-one mappings + * # because it doesn't rescale the overall result + */ + __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_sent_len); + + /* "average_inner.pyx":256 + * # If we perform the a*x on memory, the computation is compatible with many-to-one mappings + * # because it doesn't rescale the overall result + * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) # <<<<<<<<<<<<<< + * + * cdef void compute_ft_sentence_averages(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: + */ + __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&__pyx_v_13average_inner_ONE)); + + /* "average_inner.pyx":252 + * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + * + * if sent_len > ZEROF: # <<<<<<<<<<<<<< + * inv_count = ONEF / sent_len + * # If we perform the a*x on memory, the computation is compatible with many-to-one mappings + */ + } + } + + /* "average_inner.pyx":213 + * return eff_sents, eff_words + * + * cdef void compute_base_sentence_averages(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< + * """Perform optimized sentence-level averaging for BaseAny2Vec model. + * + */ + + /* function exit code */ +} + +/* "average_inner.pyx":258 + * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + * + * cdef void compute_ft_sentence_averages(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< + * """Perform optimized sentence-level averaging for FastText model. + * + */ + +static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { + int __pyx_v_size; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_end; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; + __pyx_t_13average_inner_uINT_t __pyx_v_ngram_row; + __pyx_t_13average_inner_uINT_t __pyx_v_ngrams; + __pyx_t_13average_inner_uINT_t __pyx_v_i; + __pyx_t_13average_inner_uINT_t __pyx_v_j; + __pyx_t_13average_inner_uINT_t __pyx_v_word_idx; + __pyx_t_13average_inner_uINT_t __pyx_v_word_row; + __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; + __pyx_t_13average_inner_REAL_t __pyx_v_inv_count; + __pyx_t_13average_inner_REAL_t __pyx_v_inv_ngram; + CYTHON_UNUSED __pyx_t_13average_inner_REAL_t __pyx_v_oov_weight; + int __pyx_t_1; + __pyx_t_13average_inner_REAL_t __pyx_t_2; + __pyx_t_13average_inner_uINT_t __pyx_t_3; + __pyx_t_13average_inner_uINT_t __pyx_t_4; + __pyx_t_13average_inner_uINT_t __pyx_t_5; + __pyx_t_13average_inner_uINT_t __pyx_t_6; + __pyx_t_13average_inner_uINT_t __pyx_t_7; + __pyx_t_13average_inner_uINT_t __pyx_t_8; + int __pyx_t_9; + __pyx_t_13average_inner_uINT_t __pyx_t_10; + __pyx_t_13average_inner_uINT_t __pyx_t_11; + __pyx_t_13average_inner_uINT_t __pyx_t_12; + + /* "average_inner.pyx":274 + * """ + * cdef: + * int size = c.size # <<<<<<<<<<<<<< + * + * uINT_t sent_idx, sent_start, sent_end, sent_row + */ + __pyx_t_1 = __pyx_v_c->size; + __pyx_v_size = __pyx_t_1; + + /* "average_inner.pyx":284 + * REAL_t sent_len + * REAL_t inv_count, inv_ngram + * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< + * + * + */ + __pyx_t_2 = __pyx_v_c->oov_weight; + __pyx_v_oov_weight = __pyx_t_2; + + /* "average_inner.pyx":287 + * + * + * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * sent_start = c.sentence_boundary[sent_idx] + */ + __pyx_t_3 = __pyx_v_num_sentences; + __pyx_t_4 = __pyx_t_3; + for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { + __pyx_v_sent_idx = __pyx_t_5; + + /* "average_inner.pyx":288 + * + * for sent_idx in range(num_sentences): + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] + */ + (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); + + /* "average_inner.pyx":289 + * for sent_idx in range(num_sentences): + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< + * sent_end = c.sentence_boundary[sent_idx + 1] + * sent_len = ZEROF + */ + __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); + + /* "average_inner.pyx":290 + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< + * sent_len = ZEROF + * + */ + __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); + + /* "average_inner.pyx":291 + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] + * sent_len = ZEROF # <<<<<<<<<<<<<< + * + * for i in range(sent_start, sent_end): + */ + __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; + + /* "average_inner.pyx":293 + * sent_len = ZEROF + * + * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< + * sent_len += ONEF + * sent_row = c.sent_adresses[i] * size + */ + __pyx_t_6 = __pyx_v_sent_end; + __pyx_t_7 = __pyx_t_6; + for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { + __pyx_v_i = __pyx_t_8; + + /* "average_inner.pyx":294 + * + * for i in range(sent_start, sent_end): + * sent_len += ONEF # <<<<<<<<<<<<<< + * sent_row = c.sent_adresses[i] * size + * + */ + __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); + + /* "average_inner.pyx":295 + * for i in range(sent_start, sent_end): + * sent_len += ONEF + * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< + * + * word_idx = c.word_indices[i] + */ + __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); + + /* "average_inner.pyx":297 + * sent_row = c.sent_adresses[i] * size + * + * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< + * ngrams = c.subwords_idx_len[i] + * + */ + __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); + + /* "average_inner.pyx":298 + * + * word_idx = c.word_indices[i] + * ngrams = c.subwords_idx_len[i] # <<<<<<<<<<<<<< + * + * if ngrams == 0: + */ + __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_i]); + + /* "average_inner.pyx":300 + * ngrams = c.subwords_idx_len[i] + * + * if ngrams == 0: # <<<<<<<<<<<<<< + * word_row = c.word_indices[i] * size + * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + */ + __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); + if (__pyx_t_9) { + + /* "average_inner.pyx":301 + * + * if ngrams == 0: + * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< + * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + * else: + */ + __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); + + /* "average_inner.pyx":302 + * if ngrams == 0: + * word_row = c.word_indices[i] * size + * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) # <<<<<<<<<<<<<< + * else: + * inv_ngram = (ONEF / ngrams) * c.oov_weight + */ + __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); + + /* "average_inner.pyx":300 + * ngrams = c.subwords_idx_len[i] + * + * if ngrams == 0: # <<<<<<<<<<<<<< + * word_row = c.word_indices[i] * size + * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + */ + goto __pyx_L7; + } + + /* "average_inner.pyx":304 + * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + * else: + * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< + * for j in range(ngrams): + * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size + */ + /*else*/ { + __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); + + /* "average_inner.pyx":305 + * else: + * inv_ngram = (ONEF / ngrams) * c.oov_weight + * for j in range(ngrams): # <<<<<<<<<<<<<< + * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size + * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) + */ + __pyx_t_10 = __pyx_v_ngrams; + __pyx_t_11 = __pyx_t_10; + for (__pyx_t_12 = 0; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { + __pyx_v_j = __pyx_t_12; + + /* "average_inner.pyx":306 + * inv_ngram = (ONEF / ngrams) * c.oov_weight + * for j in range(ngrams): + * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size # <<<<<<<<<<<<<< + * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) + * + */ + __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_i * 40) + __pyx_v_j)]) * __pyx_v_size); + + /* "average_inner.pyx":307 + * for j in range(ngrams): + * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size + * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) # <<<<<<<<<<<<<< + * + * if sent_len > ZEROF: + */ + __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_ngram), (&(__pyx_v_c->ngram_vectors[__pyx_v_ngram_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); + } + } + __pyx_L7:; + } + + /* "average_inner.pyx":309 + * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) + * + * if sent_len > ZEROF: # <<<<<<<<<<<<<< + * inv_count = ONEF / sent_len + * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + */ + __pyx_t_9 = ((__pyx_v_sent_len > __pyx_v_13average_inner_ZEROF) != 0); + if (__pyx_t_9) { + + /* "average_inner.pyx":310 + * + * if sent_len > ZEROF: + * inv_count = ONEF / sent_len # <<<<<<<<<<<<<< + * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + * + */ + __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_sent_len); + + /* "average_inner.pyx":311 + * if sent_len > ZEROF: + * inv_count = ONEF / sent_len + * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) # <<<<<<<<<<<<<< + * + * def train_average_cy(model, indexed_sentences, target, memory): + */ + __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&__pyx_v_13average_inner_ONE)); + + /* "average_inner.pyx":309 + * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) + * + * if sent_len > ZEROF: # <<<<<<<<<<<<<< + * inv_count = ONEF / sent_len + * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + */ + } + } + + /* "average_inner.pyx":258 + * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + * + * cdef void compute_ft_sentence_averages(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< + * """Perform optimized sentence-level averaging for FastText model. + * + */ + + /* function exit code */ +} + +/* "average_inner.pyx":313 + * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + * + * def train_average_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< + * """Training on a sequence of sentences and update the target ndarray. + * + */ + +/* Python wrapper */ +static PyObject *__pyx_pw_13average_inner_1train_average_cy(PyObject *__pyx_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/ +static char __pyx_doc_13average_inner_train_average_cy[] = "train_average_cy(model, indexed_sentences, target, memory)\nTraining on a sequence of sentences and update the target ndarray.\n\n Called internally from :meth:`~fse.models.average.Average._do_train_job`.\n\n Parameters\n ----------\n model : :class:`~fse.models.base_s2v.BaseSentence2VecModel`\n The BaseSentence2VecModel model instance.\n indexed_sentences : iterable of tuple\n The sentences used to train the model.\n target : ndarray\n The target ndarray. We use the index from indexed_sentences\n to write into the corresponding row of target.\n memory : ndarray\n Private memory for each working thread.\n\n Returns\n -------\n int, int\n Number of effective sentences (non-zero) and effective words in the vocabulary used \n during training the sentence embedding.\n "; +static PyMethodDef __pyx_mdef_13average_inner_1train_average_cy = {"train_average_cy", (PyCFunction)(void*)(PyCFunctionWithKeywords)__pyx_pw_13average_inner_1train_average_cy, METH_VARARGS|METH_KEYWORDS, __pyx_doc_13average_inner_train_average_cy}; +static PyObject *__pyx_pw_13average_inner_1train_average_cy(PyObject *__pyx_self, PyObject *__pyx_args, PyObject *__pyx_kwds) { + PyObject *__pyx_v_model = 0; + PyObject *__pyx_v_indexed_sentences = 0; + PyObject *__pyx_v_target = 0; + PyObject *__pyx_v_memory = 0; + PyObject *__pyx_r = 0; + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("train_average_cy (wrapper)", 0); + { + static PyObject **__pyx_pyargnames[] = {&__pyx_n_s_model,&__pyx_n_s_indexed_sentences,&__pyx_n_s_target,&__pyx_n_s_memory,0}; + PyObject* values[4] = {0,0,0,0}; + if (unlikely(__pyx_kwds)) { + Py_ssize_t kw_args; + const Py_ssize_t pos_args = PyTuple_GET_SIZE(__pyx_args); + switch (pos_args) { + case 4: values[3] = PyTuple_GET_ITEM(__pyx_args, 3); + CYTHON_FALLTHROUGH; + case 3: values[2] = PyTuple_GET_ITEM(__pyx_args, 2); + CYTHON_FALLTHROUGH; + case 2: values[1] = PyTuple_GET_ITEM(__pyx_args, 1); + CYTHON_FALLTHROUGH; + case 1: values[0] = PyTuple_GET_ITEM(__pyx_args, 0); + CYTHON_FALLTHROUGH; + case 0: break; + default: goto __pyx_L5_argtuple_error; + } + kw_args = PyDict_Size(__pyx_kwds); + switch (pos_args) { + case 0: + if (likely((values[0] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_model)) != 0)) kw_args--; + else goto __pyx_L5_argtuple_error; + CYTHON_FALLTHROUGH; + case 1: + if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; + else { + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 1); __PYX_ERR(0, 313, __pyx_L3_error) + } + CYTHON_FALLTHROUGH; + case 2: + if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; + else { + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 2); __PYX_ERR(0, 313, __pyx_L3_error) + } + CYTHON_FALLTHROUGH; + case 3: + if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; + else { + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 3); __PYX_ERR(0, 313, __pyx_L3_error) + } + } + if (unlikely(kw_args > 0)) { + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_average_cy") < 0)) __PYX_ERR(0, 313, __pyx_L3_error) + } + } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { + goto __pyx_L5_argtuple_error; + } else { + values[0] = PyTuple_GET_ITEM(__pyx_args, 0); + values[1] = PyTuple_GET_ITEM(__pyx_args, 1); + values[2] = PyTuple_GET_ITEM(__pyx_args, 2); + values[3] = PyTuple_GET_ITEM(__pyx_args, 3); + } + __pyx_v_model = values[0]; + __pyx_v_indexed_sentences = values[1]; + __pyx_v_target = values[2]; + __pyx_v_memory = values[3]; + } + goto __pyx_L4_argument_unpacking_done; + __pyx_L5_argtuple_error:; + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 313, __pyx_L3_error) + __pyx_L3_error:; + __Pyx_AddTraceback("average_inner.train_average_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); + __Pyx_RefNannyFinishContext(); + return NULL; + __pyx_L4_argument_unpacking_done:; + __pyx_r = __pyx_pf_13average_inner_train_average_cy(__pyx_self, __pyx_v_model, __pyx_v_indexed_sentences, __pyx_v_target, __pyx_v_memory); + + /* function exit code */ + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObject *__pyx_self, PyObject *__pyx_v_model, PyObject *__pyx_v_indexed_sentences, PyObject *__pyx_v_target, PyObject *__pyx_v_memory) { + __pyx_t_13average_inner_uINT_t __pyx_v_eff_sentences; + __pyx_t_13average_inner_uINT_t __pyx_v_eff_words; + struct __pyx_t_13average_inner_BaseSentenceVecsConfig __pyx_v_w2v; + struct __pyx_t_13average_inner_FTSentenceVecsConfig __pyx_v_ft; + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + int __pyx_t_2; + int __pyx_t_3; + PyObject *__pyx_t_4 = NULL; + PyObject *__pyx_t_5 = NULL; + PyObject *__pyx_t_6 = NULL; + PyObject *(*__pyx_t_7)(PyObject *); + __pyx_t_13average_inner_uINT_t __pyx_t_8; + __pyx_t_13average_inner_uINT_t __pyx_t_9; + __Pyx_RefNannySetupContext("train_average_cy", 0); + + /* "average_inner.pyx":337 + * """ + * + * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< + * cdef uINT_t eff_words = 0 + * cdef BaseSentenceVecsConfig w2v + */ + __pyx_v_eff_sentences = 0; + + /* "average_inner.pyx":338 + * + * cdef uINT_t eff_sentences = 0 + * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< + * cdef BaseSentenceVecsConfig w2v + * cdef FTSentenceVecsConfig ft + */ + __pyx_v_eff_words = 0; + + /* "average_inner.pyx":342 + * cdef FTSentenceVecsConfig ft + * + * if not model.is_ft: # <<<<<<<<<<<<<< + * init_base_s2v_config(&w2v, model, target, memory) + * + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 342, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 342, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_3 = ((!__pyx_t_2) != 0); + if (__pyx_t_3) { + + /* "average_inner.pyx":343 + * + * if not model.is_ft: + * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< + * + * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + */ + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 343, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":345 + * init_base_s2v_config(&w2v, model, target, memory) + * + * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) # <<<<<<<<<<<<<< + * + * with nogil: + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 345, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 345, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 345, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { + PyObject* sequence = __pyx_t_1; + Py_ssize_t size = __Pyx_PySequence_SIZE(sequence); + if (unlikely(size != 2)) { + if (size > 2) __Pyx_RaiseTooManyValuesError(2); + else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); + __PYX_ERR(0, 345, __pyx_L1_error) + } + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + if (likely(PyTuple_CheckExact(sequence))) { + __pyx_t_4 = PyTuple_GET_ITEM(sequence, 0); + __pyx_t_5 = PyTuple_GET_ITEM(sequence, 1); + } else { + __pyx_t_4 = PyList_GET_ITEM(sequence, 0); + __pyx_t_5 = PyList_GET_ITEM(sequence, 1); + } + __Pyx_INCREF(__pyx_t_4); + __Pyx_INCREF(__pyx_t_5); + #else + __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 345, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 345, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_5); + #endif + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + } else { + Py_ssize_t index = -1; + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 345, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_6); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; + index = 0; __pyx_t_4 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_4)) goto __pyx_L4_unpacking_failed; + __Pyx_GOTREF(__pyx_t_4); + index = 1; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L4_unpacking_failed; + __Pyx_GOTREF(__pyx_t_5); + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 345, __pyx_L1_error) + __pyx_t_7 = NULL; + __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; + goto __pyx_L5_unpacking_done; + __pyx_L4_unpacking_failed:; + __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; + __pyx_t_7 = NULL; + if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); + __PYX_ERR(0, 345, __pyx_L1_error) + __pyx_L5_unpacking_done:; + } + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 345, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 345, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; + __pyx_v_eff_sentences = __pyx_t_8; + __pyx_v_eff_words = __pyx_t_9; + + /* "average_inner.pyx":347 + * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + * + * with nogil: # <<<<<<<<<<<<<< + * compute_base_sentence_averages(&w2v, eff_sentences) + * else: + */ + { + #ifdef WITH_THREAD + PyThreadState *_save; + Py_UNBLOCK_THREADS + __Pyx_FastGIL_Remember(); + #endif + /*try:*/ { + + /* "average_inner.pyx":348 + * + * with nogil: + * compute_base_sentence_averages(&w2v, eff_sentences) # <<<<<<<<<<<<<< + * else: + * init_ft_s2v_config(&ft, model, target, memory) + */ + __pyx_f_13average_inner_compute_base_sentence_averages((&__pyx_v_w2v), __pyx_v_eff_sentences); + } + + /* "average_inner.pyx":347 + * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + * + * with nogil: # <<<<<<<<<<<<<< + * compute_base_sentence_averages(&w2v, eff_sentences) + * else: + */ + /*finally:*/ { + /*normal exit:*/{ + #ifdef WITH_THREAD + __Pyx_FastGIL_Forget(); + Py_BLOCK_THREADS + #endif + goto __pyx_L8; + } + __pyx_L8:; + } + } + + /* "average_inner.pyx":342 + * cdef FTSentenceVecsConfig ft + * + * if not model.is_ft: # <<<<<<<<<<<<<< + * init_base_s2v_config(&w2v, model, target, memory) + * + */ + goto __pyx_L3; + } + + /* "average_inner.pyx":350 + * compute_base_sentence_averages(&w2v, eff_sentences) + * else: + * init_ft_s2v_config(&ft, model, target, memory) # <<<<<<<<<<<<<< + * + * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + */ + /*else*/ { + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 350, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":352 + * init_ft_s2v_config(&ft, model, target, memory) + * + * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) # <<<<<<<<<<<<<< + * + * with nogil: + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 352, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 352, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_5); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 352, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; + if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { + PyObject* sequence = __pyx_t_1; + Py_ssize_t size = __Pyx_PySequence_SIZE(sequence); + if (unlikely(size != 2)) { + if (size > 2) __Pyx_RaiseTooManyValuesError(2); + else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); + __PYX_ERR(0, 352, __pyx_L1_error) + } + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + if (likely(PyTuple_CheckExact(sequence))) { + __pyx_t_5 = PyTuple_GET_ITEM(sequence, 0); + __pyx_t_4 = PyTuple_GET_ITEM(sequence, 1); + } else { + __pyx_t_5 = PyList_GET_ITEM(sequence, 0); + __pyx_t_4 = PyList_GET_ITEM(sequence, 1); + } + __Pyx_INCREF(__pyx_t_5); + __Pyx_INCREF(__pyx_t_4); + #else + __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 352, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_5); + __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 352, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + #endif + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + } else { + Py_ssize_t index = -1; + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 352, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_6); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; + index = 0; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L9_unpacking_failed; + __Pyx_GOTREF(__pyx_t_5); + index = 1; __pyx_t_4 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_4)) goto __pyx_L9_unpacking_failed; + __Pyx_GOTREF(__pyx_t_4); + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 352, __pyx_L1_error) + __pyx_t_7 = NULL; + __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; + goto __pyx_L10_unpacking_done; + __pyx_L9_unpacking_failed:; + __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; + __pyx_t_7 = NULL; + if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); + __PYX_ERR(0, 352, __pyx_L1_error) + __pyx_L10_unpacking_done:; + } + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 352, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 352, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_v_eff_sentences = __pyx_t_9; + __pyx_v_eff_words = __pyx_t_8; + + /* "average_inner.pyx":354 + * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + * + * with nogil: # <<<<<<<<<<<<<< + * compute_ft_sentence_averages(&ft, eff_sentences) + * + */ + { + #ifdef WITH_THREAD + PyThreadState *_save; + Py_UNBLOCK_THREADS + __Pyx_FastGIL_Remember(); + #endif + /*try:*/ { + + /* "average_inner.pyx":355 + * + * with nogil: + * compute_ft_sentence_averages(&ft, eff_sentences) # <<<<<<<<<<<<<< + * + * return eff_sentences, eff_words + */ + __pyx_f_13average_inner_compute_ft_sentence_averages((&__pyx_v_ft), __pyx_v_eff_sentences); + } + + /* "average_inner.pyx":354 + * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + * + * with nogil: # <<<<<<<<<<<<<< + * compute_ft_sentence_averages(&ft, eff_sentences) + * + */ + /*finally:*/ { + /*normal exit:*/{ + #ifdef WITH_THREAD + __Pyx_FastGIL_Forget(); + Py_BLOCK_THREADS + #endif + goto __pyx_L13; + } + __pyx_L13:; + } + } + } + __pyx_L3:; + + /* "average_inner.pyx":357 + * compute_ft_sentence_averages(&ft, eff_sentences) + * + * return eff_sentences, eff_words # <<<<<<<<<<<<<< + * + * def init(): + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 357, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 357, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 357, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_5); + __Pyx_GIVEREF(__pyx_t_1); + PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_1); + __Pyx_GIVEREF(__pyx_t_4); + PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_4); + __pyx_t_1 = 0; + __pyx_t_4 = 0; + __pyx_r = __pyx_t_5; + __pyx_t_5 = 0; + goto __pyx_L0; + + /* "average_inner.pyx":313 + * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + * + * def train_average_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< + * """Training on a sequence of sentences and update the target ndarray. + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_XDECREF(__pyx_t_4); + __Pyx_XDECREF(__pyx_t_5); + __Pyx_XDECREF(__pyx_t_6); + __Pyx_AddTraceback("average_inner.train_average_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = NULL; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "average_inner.pyx":359 + * return eff_sentences, eff_words + * + * def init(): # <<<<<<<<<<<<<< + * return 1 + * + */ + +/* Python wrapper */ +static PyObject *__pyx_pw_13average_inner_3init(PyObject *__pyx_self, CYTHON_UNUSED PyObject *unused); /*proto*/ +static char __pyx_doc_13average_inner_2init[] = "init()"; +static PyMethodDef __pyx_mdef_13average_inner_3init = {"init", (PyCFunction)__pyx_pw_13average_inner_3init, METH_NOARGS, __pyx_doc_13average_inner_2init}; +static PyObject *__pyx_pw_13average_inner_3init(PyObject *__pyx_self, CYTHON_UNUSED PyObject *unused) { + PyObject *__pyx_r = 0; + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("init (wrapper)", 0); + __pyx_r = __pyx_pf_13average_inner_2init(__pyx_self); + + /* function exit code */ + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +static PyObject *__pyx_pf_13average_inner_2init(CYTHON_UNUSED PyObject *__pyx_self) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("init", 0); + + /* "average_inner.pyx":360 + * + * def init(): + * return 1 # <<<<<<<<<<<<<< + * + * MAX_WORDS_IN_BATCH = MAX_WORDS + */ + __Pyx_XDECREF(__pyx_r); + __Pyx_INCREF(__pyx_int_1); + __pyx_r = __pyx_int_1; + goto __pyx_L0; + + /* "average_inner.pyx":359 + * return eff_sentences, eff_words + * + * def init(): # <<<<<<<<<<<<<< + * return 1 + * + */ + + /* function exit code */ + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":258 + * # experimental exception made for __getbuffer__ and __releasebuffer__ + * # -- the details of this may change. + * def __getbuffer__(ndarray self, Py_buffer* info, int flags): # <<<<<<<<<<<<<< + * # This implementation of getbuffer is geared towards Cython + * # requirements, and does not yet fulfill the PEP. + */ + +/* Python wrapper */ +static CYTHON_UNUSED int __pyx_pw_5numpy_7ndarray_1__getbuffer__(PyObject *__pyx_v_self, Py_buffer *__pyx_v_info, int __pyx_v_flags); /*proto*/ +static CYTHON_UNUSED int __pyx_pw_5numpy_7ndarray_1__getbuffer__(PyObject *__pyx_v_self, Py_buffer *__pyx_v_info, int __pyx_v_flags) { + int __pyx_r; + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__getbuffer__ (wrapper)", 0); + __pyx_r = __pyx_pf_5numpy_7ndarray___getbuffer__(((PyArrayObject *)__pyx_v_self), ((Py_buffer *)__pyx_v_info), ((int)__pyx_v_flags)); + + /* function exit code */ + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +static int __pyx_pf_5numpy_7ndarray___getbuffer__(PyArrayObject *__pyx_v_self, Py_buffer *__pyx_v_info, int __pyx_v_flags) { + int __pyx_v_i; + int __pyx_v_ndim; + int __pyx_v_endian_detector; + int __pyx_v_little_endian; + int __pyx_v_t; + char *__pyx_v_f; + PyArray_Descr *__pyx_v_descr = 0; + int __pyx_v_offset; + int __pyx_r; + __Pyx_RefNannyDeclarations + int __pyx_t_1; + int __pyx_t_2; + PyObject *__pyx_t_3 = NULL; + int __pyx_t_4; + int __pyx_t_5; + int __pyx_t_6; + PyArray_Descr *__pyx_t_7; + PyObject *__pyx_t_8 = NULL; + char *__pyx_t_9; + if (__pyx_v_info == NULL) { + PyErr_SetString(PyExc_BufferError, "PyObject_GetBuffer: view==NULL argument is obsolete"); + return -1; + } + __Pyx_RefNannySetupContext("__getbuffer__", 0); + __pyx_v_info->obj = Py_None; __Pyx_INCREF(Py_None); + __Pyx_GIVEREF(__pyx_v_info->obj); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":265 + * + * cdef int i, ndim + * cdef int endian_detector = 1 # <<<<<<<<<<<<<< + * cdef bint little_endian = ((&endian_detector)[0] != 0) + * + */ + __pyx_v_endian_detector = 1; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":266 + * cdef int i, ndim + * cdef int endian_detector = 1 + * cdef bint little_endian = ((&endian_detector)[0] != 0) # <<<<<<<<<<<<<< + * + * ndim = PyArray_NDIM(self) + */ + __pyx_v_little_endian = ((((char *)(&__pyx_v_endian_detector))[0]) != 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":268 + * cdef bint little_endian = ((&endian_detector)[0] != 0) + * + * ndim = PyArray_NDIM(self) # <<<<<<<<<<<<<< + * + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) + */ + __pyx_v_ndim = PyArray_NDIM(__pyx_v_self); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":270 + * ndim = PyArray_NDIM(self) + * + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) # <<<<<<<<<<<<<< + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_C_CONTIGUOUS)): + * raise ValueError(u"ndarray is not C contiguous") + */ + __pyx_t_2 = (((__pyx_v_flags & PyBUF_C_CONTIGUOUS) == PyBUF_C_CONTIGUOUS) != 0); + if (__pyx_t_2) { + } else { + __pyx_t_1 = __pyx_t_2; + goto __pyx_L4_bool_binop_done; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":271 + * + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_C_CONTIGUOUS)): # <<<<<<<<<<<<<< + * raise ValueError(u"ndarray is not C contiguous") + * + */ + __pyx_t_2 = ((!(PyArray_CHKFLAGS(__pyx_v_self, NPY_ARRAY_C_CONTIGUOUS) != 0)) != 0); + __pyx_t_1 = __pyx_t_2; + __pyx_L4_bool_binop_done:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":270 + * ndim = PyArray_NDIM(self) + * + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) # <<<<<<<<<<<<<< + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_C_CONTIGUOUS)): + * raise ValueError(u"ndarray is not C contiguous") + */ + if (unlikely(__pyx_t_1)) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":272 + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_C_CONTIGUOUS)): + * raise ValueError(u"ndarray is not C contiguous") # <<<<<<<<<<<<<< + * + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) + */ + __pyx_t_3 = __Pyx_PyObject_Call(__pyx_builtin_ValueError, __pyx_tuple_, NULL); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 272, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_Raise(__pyx_t_3, 0, 0, 0); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __PYX_ERR(1, 272, __pyx_L1_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":270 + * ndim = PyArray_NDIM(self) + * + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) # <<<<<<<<<<<<<< + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_C_CONTIGUOUS)): + * raise ValueError(u"ndarray is not C contiguous") + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":274 + * raise ValueError(u"ndarray is not C contiguous") + * + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) # <<<<<<<<<<<<<< + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_F_CONTIGUOUS)): + * raise ValueError(u"ndarray is not Fortran contiguous") + */ + __pyx_t_2 = (((__pyx_v_flags & PyBUF_F_CONTIGUOUS) == PyBUF_F_CONTIGUOUS) != 0); + if (__pyx_t_2) { + } else { + __pyx_t_1 = __pyx_t_2; + goto __pyx_L7_bool_binop_done; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":275 + * + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_F_CONTIGUOUS)): # <<<<<<<<<<<<<< + * raise ValueError(u"ndarray is not Fortran contiguous") + * + */ + __pyx_t_2 = ((!(PyArray_CHKFLAGS(__pyx_v_self, NPY_ARRAY_F_CONTIGUOUS) != 0)) != 0); + __pyx_t_1 = __pyx_t_2; + __pyx_L7_bool_binop_done:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":274 + * raise ValueError(u"ndarray is not C contiguous") + * + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) # <<<<<<<<<<<<<< + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_F_CONTIGUOUS)): + * raise ValueError(u"ndarray is not Fortran contiguous") + */ + if (unlikely(__pyx_t_1)) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":276 + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_F_CONTIGUOUS)): + * raise ValueError(u"ndarray is not Fortran contiguous") # <<<<<<<<<<<<<< + * + * info.buf = PyArray_DATA(self) + */ + __pyx_t_3 = __Pyx_PyObject_Call(__pyx_builtin_ValueError, __pyx_tuple__2, NULL); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 276, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_Raise(__pyx_t_3, 0, 0, 0); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __PYX_ERR(1, 276, __pyx_L1_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":274 + * raise ValueError(u"ndarray is not C contiguous") + * + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) # <<<<<<<<<<<<<< + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_F_CONTIGUOUS)): + * raise ValueError(u"ndarray is not Fortran contiguous") + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":278 + * raise ValueError(u"ndarray is not Fortran contiguous") + * + * info.buf = PyArray_DATA(self) # <<<<<<<<<<<<<< + * info.ndim = ndim + * if sizeof(npy_intp) != sizeof(Py_ssize_t): + */ + __pyx_v_info->buf = PyArray_DATA(__pyx_v_self); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":279 + * + * info.buf = PyArray_DATA(self) + * info.ndim = ndim # <<<<<<<<<<<<<< + * if sizeof(npy_intp) != sizeof(Py_ssize_t): + * # Allocate new buffer for strides and shape info. + */ + __pyx_v_info->ndim = __pyx_v_ndim; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":280 + * info.buf = PyArray_DATA(self) + * info.ndim = ndim + * if sizeof(npy_intp) != sizeof(Py_ssize_t): # <<<<<<<<<<<<<< + * # Allocate new buffer for strides and shape info. + * # This is allocated as one block, strides first. + */ + __pyx_t_1 = (((sizeof(npy_intp)) != (sizeof(Py_ssize_t))) != 0); + if (__pyx_t_1) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":283 + * # Allocate new buffer for strides and shape info. + * # This is allocated as one block, strides first. + * info.strides = PyObject_Malloc(sizeof(Py_ssize_t) * 2 * ndim) # <<<<<<<<<<<<<< + * info.shape = info.strides + ndim + * for i in range(ndim): + */ + __pyx_v_info->strides = ((Py_ssize_t *)PyObject_Malloc((((sizeof(Py_ssize_t)) * 2) * ((size_t)__pyx_v_ndim)))); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":284 + * # This is allocated as one block, strides first. + * info.strides = PyObject_Malloc(sizeof(Py_ssize_t) * 2 * ndim) + * info.shape = info.strides + ndim # <<<<<<<<<<<<<< + * for i in range(ndim): + * info.strides[i] = PyArray_STRIDES(self)[i] + */ + __pyx_v_info->shape = (__pyx_v_info->strides + __pyx_v_ndim); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":285 + * info.strides = PyObject_Malloc(sizeof(Py_ssize_t) * 2 * ndim) + * info.shape = info.strides + ndim + * for i in range(ndim): # <<<<<<<<<<<<<< + * info.strides[i] = PyArray_STRIDES(self)[i] + * info.shape[i] = PyArray_DIMS(self)[i] + */ + __pyx_t_4 = __pyx_v_ndim; + __pyx_t_5 = __pyx_t_4; + for (__pyx_t_6 = 0; __pyx_t_6 < __pyx_t_5; __pyx_t_6+=1) { + __pyx_v_i = __pyx_t_6; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":286 + * info.shape = info.strides + ndim + * for i in range(ndim): + * info.strides[i] = PyArray_STRIDES(self)[i] # <<<<<<<<<<<<<< + * info.shape[i] = PyArray_DIMS(self)[i] + * else: + */ + (__pyx_v_info->strides[__pyx_v_i]) = (PyArray_STRIDES(__pyx_v_self)[__pyx_v_i]); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":287 + * for i in range(ndim): + * info.strides[i] = PyArray_STRIDES(self)[i] + * info.shape[i] = PyArray_DIMS(self)[i] # <<<<<<<<<<<<<< + * else: + * info.strides = PyArray_STRIDES(self) + */ + (__pyx_v_info->shape[__pyx_v_i]) = (PyArray_DIMS(__pyx_v_self)[__pyx_v_i]); + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":280 + * info.buf = PyArray_DATA(self) + * info.ndim = ndim + * if sizeof(npy_intp) != sizeof(Py_ssize_t): # <<<<<<<<<<<<<< + * # Allocate new buffer for strides and shape info. + * # This is allocated as one block, strides first. + */ + goto __pyx_L9; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":289 + * info.shape[i] = PyArray_DIMS(self)[i] + * else: + * info.strides = PyArray_STRIDES(self) # <<<<<<<<<<<<<< + * info.shape = PyArray_DIMS(self) + * info.suboffsets = NULL + */ + /*else*/ { + __pyx_v_info->strides = ((Py_ssize_t *)PyArray_STRIDES(__pyx_v_self)); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":290 + * else: + * info.strides = PyArray_STRIDES(self) + * info.shape = PyArray_DIMS(self) # <<<<<<<<<<<<<< + * info.suboffsets = NULL + * info.itemsize = PyArray_ITEMSIZE(self) + */ + __pyx_v_info->shape = ((Py_ssize_t *)PyArray_DIMS(__pyx_v_self)); + } + __pyx_L9:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":291 + * info.strides = PyArray_STRIDES(self) + * info.shape = PyArray_DIMS(self) + * info.suboffsets = NULL # <<<<<<<<<<<<<< + * info.itemsize = PyArray_ITEMSIZE(self) + * info.readonly = not PyArray_ISWRITEABLE(self) + */ + __pyx_v_info->suboffsets = NULL; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":292 + * info.shape = PyArray_DIMS(self) + * info.suboffsets = NULL + * info.itemsize = PyArray_ITEMSIZE(self) # <<<<<<<<<<<<<< + * info.readonly = not PyArray_ISWRITEABLE(self) + * + */ + __pyx_v_info->itemsize = PyArray_ITEMSIZE(__pyx_v_self); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":293 + * info.suboffsets = NULL + * info.itemsize = PyArray_ITEMSIZE(self) + * info.readonly = not PyArray_ISWRITEABLE(self) # <<<<<<<<<<<<<< + * + * cdef int t + */ + __pyx_v_info->readonly = (!(PyArray_ISWRITEABLE(__pyx_v_self) != 0)); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":296 + * + * cdef int t + * cdef char* f = NULL # <<<<<<<<<<<<<< + * cdef dtype descr = PyArray_DESCR(self) + * cdef int offset + */ + __pyx_v_f = NULL; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":297 + * cdef int t + * cdef char* f = NULL + * cdef dtype descr = PyArray_DESCR(self) # <<<<<<<<<<<<<< + * cdef int offset + * + */ + __pyx_t_7 = PyArray_DESCR(__pyx_v_self); + __pyx_t_3 = ((PyObject *)__pyx_t_7); + __Pyx_INCREF(__pyx_t_3); + __pyx_v_descr = ((PyArray_Descr *)__pyx_t_3); + __pyx_t_3 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":300 + * cdef int offset + * + * info.obj = self # <<<<<<<<<<<<<< + * + * if not PyDataType_HASFIELDS(descr): + */ + __Pyx_INCREF(((PyObject *)__pyx_v_self)); + __Pyx_GIVEREF(((PyObject *)__pyx_v_self)); + __Pyx_GOTREF(__pyx_v_info->obj); + __Pyx_DECREF(__pyx_v_info->obj); + __pyx_v_info->obj = ((PyObject *)__pyx_v_self); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":302 + * info.obj = self + * + * if not PyDataType_HASFIELDS(descr): # <<<<<<<<<<<<<< + * t = descr.type_num + * if ((descr.byteorder == c'>' and little_endian) or + */ + __pyx_t_1 = ((!(PyDataType_HASFIELDS(__pyx_v_descr) != 0)) != 0); + if (__pyx_t_1) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":303 + * + * if not PyDataType_HASFIELDS(descr): + * t = descr.type_num # <<<<<<<<<<<<<< + * if ((descr.byteorder == c'>' and little_endian) or + * (descr.byteorder == c'<' and not little_endian)): + */ + __pyx_t_4 = __pyx_v_descr->type_num; + __pyx_v_t = __pyx_t_4; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":304 + * if not PyDataType_HASFIELDS(descr): + * t = descr.type_num + * if ((descr.byteorder == c'>' and little_endian) or # <<<<<<<<<<<<<< + * (descr.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + */ + __pyx_t_2 = ((__pyx_v_descr->byteorder == '>') != 0); + if (!__pyx_t_2) { + goto __pyx_L15_next_or; + } else { + } + __pyx_t_2 = (__pyx_v_little_endian != 0); + if (!__pyx_t_2) { + } else { + __pyx_t_1 = __pyx_t_2; + goto __pyx_L14_bool_binop_done; + } + __pyx_L15_next_or:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":305 + * t = descr.type_num + * if ((descr.byteorder == c'>' and little_endian) or + * (descr.byteorder == c'<' and not little_endian)): # <<<<<<<<<<<<<< + * raise ValueError(u"Non-native byte order not supported") + * if t == NPY_BYTE: f = "b" + */ + __pyx_t_2 = ((__pyx_v_descr->byteorder == '<') != 0); + if (__pyx_t_2) { + } else { + __pyx_t_1 = __pyx_t_2; + goto __pyx_L14_bool_binop_done; + } + __pyx_t_2 = ((!(__pyx_v_little_endian != 0)) != 0); + __pyx_t_1 = __pyx_t_2; + __pyx_L14_bool_binop_done:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":304 + * if not PyDataType_HASFIELDS(descr): + * t = descr.type_num + * if ((descr.byteorder == c'>' and little_endian) or # <<<<<<<<<<<<<< + * (descr.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + */ + if (unlikely(__pyx_t_1)) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":306 + * if ((descr.byteorder == c'>' and little_endian) or + * (descr.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") # <<<<<<<<<<<<<< + * if t == NPY_BYTE: f = "b" + * elif t == NPY_UBYTE: f = "B" + */ + __pyx_t_3 = __Pyx_PyObject_Call(__pyx_builtin_ValueError, __pyx_tuple__3, NULL); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 306, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_Raise(__pyx_t_3, 0, 0, 0); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __PYX_ERR(1, 306, __pyx_L1_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":304 + * if not PyDataType_HASFIELDS(descr): + * t = descr.type_num + * if ((descr.byteorder == c'>' and little_endian) or # <<<<<<<<<<<<<< + * (descr.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":307 + * (descr.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + * if t == NPY_BYTE: f = "b" # <<<<<<<<<<<<<< + * elif t == NPY_UBYTE: f = "B" + * elif t == NPY_SHORT: f = "h" + */ + switch (__pyx_v_t) { + case NPY_BYTE: + __pyx_v_f = ((char *)"b"); + break; + case NPY_UBYTE: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":308 + * raise ValueError(u"Non-native byte order not supported") + * if t == NPY_BYTE: f = "b" + * elif t == NPY_UBYTE: f = "B" # <<<<<<<<<<<<<< + * elif t == NPY_SHORT: f = "h" + * elif t == NPY_USHORT: f = "H" + */ + __pyx_v_f = ((char *)"B"); + break; + case NPY_SHORT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":309 + * if t == NPY_BYTE: f = "b" + * elif t == NPY_UBYTE: f = "B" + * elif t == NPY_SHORT: f = "h" # <<<<<<<<<<<<<< + * elif t == NPY_USHORT: f = "H" + * elif t == NPY_INT: f = "i" + */ + __pyx_v_f = ((char *)"h"); + break; + case NPY_USHORT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":310 + * elif t == NPY_UBYTE: f = "B" + * elif t == NPY_SHORT: f = "h" + * elif t == NPY_USHORT: f = "H" # <<<<<<<<<<<<<< + * elif t == NPY_INT: f = "i" + * elif t == NPY_UINT: f = "I" + */ + __pyx_v_f = ((char *)"H"); + break; + case NPY_INT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":311 + * elif t == NPY_SHORT: f = "h" + * elif t == NPY_USHORT: f = "H" + * elif t == NPY_INT: f = "i" # <<<<<<<<<<<<<< + * elif t == NPY_UINT: f = "I" + * elif t == NPY_LONG: f = "l" + */ + __pyx_v_f = ((char *)"i"); + break; + case NPY_UINT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":312 + * elif t == NPY_USHORT: f = "H" + * elif t == NPY_INT: f = "i" + * elif t == NPY_UINT: f = "I" # <<<<<<<<<<<<<< + * elif t == NPY_LONG: f = "l" + * elif t == NPY_ULONG: f = "L" + */ + __pyx_v_f = ((char *)"I"); + break; + case NPY_LONG: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":313 + * elif t == NPY_INT: f = "i" + * elif t == NPY_UINT: f = "I" + * elif t == NPY_LONG: f = "l" # <<<<<<<<<<<<<< + * elif t == NPY_ULONG: f = "L" + * elif t == NPY_LONGLONG: f = "q" + */ + __pyx_v_f = ((char *)"l"); + break; + case NPY_ULONG: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":314 + * elif t == NPY_UINT: f = "I" + * elif t == NPY_LONG: f = "l" + * elif t == NPY_ULONG: f = "L" # <<<<<<<<<<<<<< + * elif t == NPY_LONGLONG: f = "q" + * elif t == NPY_ULONGLONG: f = "Q" + */ + __pyx_v_f = ((char *)"L"); + break; + case NPY_LONGLONG: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":315 + * elif t == NPY_LONG: f = "l" + * elif t == NPY_ULONG: f = "L" + * elif t == NPY_LONGLONG: f = "q" # <<<<<<<<<<<<<< + * elif t == NPY_ULONGLONG: f = "Q" + * elif t == NPY_FLOAT: f = "f" + */ + __pyx_v_f = ((char *)"q"); + break; + case NPY_ULONGLONG: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":316 + * elif t == NPY_ULONG: f = "L" + * elif t == NPY_LONGLONG: f = "q" + * elif t == NPY_ULONGLONG: f = "Q" # <<<<<<<<<<<<<< + * elif t == NPY_FLOAT: f = "f" + * elif t == NPY_DOUBLE: f = "d" + */ + __pyx_v_f = ((char *)"Q"); + break; + case NPY_FLOAT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":317 + * elif t == NPY_LONGLONG: f = "q" + * elif t == NPY_ULONGLONG: f = "Q" + * elif t == NPY_FLOAT: f = "f" # <<<<<<<<<<<<<< + * elif t == NPY_DOUBLE: f = "d" + * elif t == NPY_LONGDOUBLE: f = "g" + */ + __pyx_v_f = ((char *)"f"); + break; + case NPY_DOUBLE: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":318 + * elif t == NPY_ULONGLONG: f = "Q" + * elif t == NPY_FLOAT: f = "f" + * elif t == NPY_DOUBLE: f = "d" # <<<<<<<<<<<<<< + * elif t == NPY_LONGDOUBLE: f = "g" + * elif t == NPY_CFLOAT: f = "Zf" + */ + __pyx_v_f = ((char *)"d"); + break; + case NPY_LONGDOUBLE: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":319 + * elif t == NPY_FLOAT: f = "f" + * elif t == NPY_DOUBLE: f = "d" + * elif t == NPY_LONGDOUBLE: f = "g" # <<<<<<<<<<<<<< + * elif t == NPY_CFLOAT: f = "Zf" + * elif t == NPY_CDOUBLE: f = "Zd" + */ + __pyx_v_f = ((char *)"g"); + break; + case NPY_CFLOAT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":320 + * elif t == NPY_DOUBLE: f = "d" + * elif t == NPY_LONGDOUBLE: f = "g" + * elif t == NPY_CFLOAT: f = "Zf" # <<<<<<<<<<<<<< + * elif t == NPY_CDOUBLE: f = "Zd" + * elif t == NPY_CLONGDOUBLE: f = "Zg" + */ + __pyx_v_f = ((char *)"Zf"); + break; + case NPY_CDOUBLE: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":321 + * elif t == NPY_LONGDOUBLE: f = "g" + * elif t == NPY_CFLOAT: f = "Zf" + * elif t == NPY_CDOUBLE: f = "Zd" # <<<<<<<<<<<<<< + * elif t == NPY_CLONGDOUBLE: f = "Zg" + * elif t == NPY_OBJECT: f = "O" + */ + __pyx_v_f = ((char *)"Zd"); + break; + case NPY_CLONGDOUBLE: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":322 + * elif t == NPY_CFLOAT: f = "Zf" + * elif t == NPY_CDOUBLE: f = "Zd" + * elif t == NPY_CLONGDOUBLE: f = "Zg" # <<<<<<<<<<<<<< + * elif t == NPY_OBJECT: f = "O" + * else: + */ + __pyx_v_f = ((char *)"Zg"); + break; + case NPY_OBJECT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":323 + * elif t == NPY_CDOUBLE: f = "Zd" + * elif t == NPY_CLONGDOUBLE: f = "Zg" + * elif t == NPY_OBJECT: f = "O" # <<<<<<<<<<<<<< + * else: + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) + */ + __pyx_v_f = ((char *)"O"); + break; + default: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":325 + * elif t == NPY_OBJECT: f = "O" + * else: + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) # <<<<<<<<<<<<<< + * info.format = f + * return + */ + __pyx_t_3 = __Pyx_PyInt_From_int(__pyx_v_t); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 325, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_8 = PyUnicode_Format(__pyx_kp_u_unknown_dtype_code_in_numpy_pxd, __pyx_t_3); if (unlikely(!__pyx_t_8)) __PYX_ERR(1, 325, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_8); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_3 = __Pyx_PyObject_CallOneArg(__pyx_builtin_ValueError, __pyx_t_8); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 325, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; + __Pyx_Raise(__pyx_t_3, 0, 0, 0); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __PYX_ERR(1, 325, __pyx_L1_error) + break; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":326 + * else: + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) + * info.format = f # <<<<<<<<<<<<<< + * return + * else: + */ + __pyx_v_info->format = __pyx_v_f; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":327 + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) + * info.format = f + * return # <<<<<<<<<<<<<< + * else: + * info.format = PyObject_Malloc(_buffer_format_string_len) + */ + __pyx_r = 0; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":302 + * info.obj = self + * + * if not PyDataType_HASFIELDS(descr): # <<<<<<<<<<<<<< + * t = descr.type_num + * if ((descr.byteorder == c'>' and little_endian) or + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":329 + * return + * else: + * info.format = PyObject_Malloc(_buffer_format_string_len) # <<<<<<<<<<<<<< + * info.format[0] = c'^' # Native data types, manual alignment + * offset = 0 + */ + /*else*/ { + __pyx_v_info->format = ((char *)PyObject_Malloc(0xFF)); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":330 + * else: + * info.format = PyObject_Malloc(_buffer_format_string_len) + * info.format[0] = c'^' # Native data types, manual alignment # <<<<<<<<<<<<<< + * offset = 0 + * f = _util_dtypestring(descr, info.format + 1, + */ + (__pyx_v_info->format[0]) = '^'; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":331 + * info.format = PyObject_Malloc(_buffer_format_string_len) + * info.format[0] = c'^' # Native data types, manual alignment + * offset = 0 # <<<<<<<<<<<<<< + * f = _util_dtypestring(descr, info.format + 1, + * info.format + _buffer_format_string_len, + */ + __pyx_v_offset = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":332 + * info.format[0] = c'^' # Native data types, manual alignment + * offset = 0 + * f = _util_dtypestring(descr, info.format + 1, # <<<<<<<<<<<<<< + * info.format + _buffer_format_string_len, + * &offset) + */ + __pyx_t_9 = __pyx_f_5numpy__util_dtypestring(__pyx_v_descr, (__pyx_v_info->format + 1), (__pyx_v_info->format + 0xFF), (&__pyx_v_offset)); if (unlikely(__pyx_t_9 == ((char *)NULL))) __PYX_ERR(1, 332, __pyx_L1_error) + __pyx_v_f = __pyx_t_9; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":335 + * info.format + _buffer_format_string_len, + * &offset) + * f[0] = c'\0' # Terminate format string # <<<<<<<<<<<<<< + * + * def __releasebuffer__(ndarray self, Py_buffer* info): + */ + (__pyx_v_f[0]) = '\x00'; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":258 + * # experimental exception made for __getbuffer__ and __releasebuffer__ + * # -- the details of this may change. + * def __getbuffer__(ndarray self, Py_buffer* info, int flags): # <<<<<<<<<<<<<< + * # This implementation of getbuffer is geared towards Cython + * # requirements, and does not yet fulfill the PEP. + */ + + /* function exit code */ + __pyx_r = 0; + goto __pyx_L0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_3); + __Pyx_XDECREF(__pyx_t_8); + __Pyx_AddTraceback("numpy.ndarray.__getbuffer__", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = -1; + if (__pyx_v_info->obj != NULL) { + __Pyx_GOTREF(__pyx_v_info->obj); + __Pyx_DECREF(__pyx_v_info->obj); __pyx_v_info->obj = 0; + } + goto __pyx_L2; + __pyx_L0:; + if (__pyx_v_info->obj == Py_None) { + __Pyx_GOTREF(__pyx_v_info->obj); + __Pyx_DECREF(__pyx_v_info->obj); __pyx_v_info->obj = 0; + } + __pyx_L2:; + __Pyx_XDECREF((PyObject *)__pyx_v_descr); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":337 + * f[0] = c'\0' # Terminate format string + * + * def __releasebuffer__(ndarray self, Py_buffer* info): # <<<<<<<<<<<<<< + * if PyArray_HASFIELDS(self): + * PyObject_Free(info.format) + */ + +/* Python wrapper */ +static CYTHON_UNUSED void __pyx_pw_5numpy_7ndarray_3__releasebuffer__(PyObject *__pyx_v_self, Py_buffer *__pyx_v_info); /*proto*/ +static CYTHON_UNUSED void __pyx_pw_5numpy_7ndarray_3__releasebuffer__(PyObject *__pyx_v_self, Py_buffer *__pyx_v_info) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__releasebuffer__ (wrapper)", 0); + __pyx_pf_5numpy_7ndarray_2__releasebuffer__(((PyArrayObject *)__pyx_v_self), ((Py_buffer *)__pyx_v_info)); + + /* function exit code */ + __Pyx_RefNannyFinishContext(); +} + +static void __pyx_pf_5numpy_7ndarray_2__releasebuffer__(PyArrayObject *__pyx_v_self, Py_buffer *__pyx_v_info) { + __Pyx_RefNannyDeclarations + int __pyx_t_1; + __Pyx_RefNannySetupContext("__releasebuffer__", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":338 + * + * def __releasebuffer__(ndarray self, Py_buffer* info): + * if PyArray_HASFIELDS(self): # <<<<<<<<<<<<<< + * PyObject_Free(info.format) + * if sizeof(npy_intp) != sizeof(Py_ssize_t): + */ + __pyx_t_1 = (PyArray_HASFIELDS(__pyx_v_self) != 0); + if (__pyx_t_1) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":339 + * def __releasebuffer__(ndarray self, Py_buffer* info): + * if PyArray_HASFIELDS(self): + * PyObject_Free(info.format) # <<<<<<<<<<<<<< + * if sizeof(npy_intp) != sizeof(Py_ssize_t): + * PyObject_Free(info.strides) + */ + PyObject_Free(__pyx_v_info->format); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":338 + * + * def __releasebuffer__(ndarray self, Py_buffer* info): + * if PyArray_HASFIELDS(self): # <<<<<<<<<<<<<< + * PyObject_Free(info.format) + * if sizeof(npy_intp) != sizeof(Py_ssize_t): + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":340 + * if PyArray_HASFIELDS(self): + * PyObject_Free(info.format) + * if sizeof(npy_intp) != sizeof(Py_ssize_t): # <<<<<<<<<<<<<< + * PyObject_Free(info.strides) + * # info.shape was stored after info.strides in the same block + */ + __pyx_t_1 = (((sizeof(npy_intp)) != (sizeof(Py_ssize_t))) != 0); + if (__pyx_t_1) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":341 + * PyObject_Free(info.format) + * if sizeof(npy_intp) != sizeof(Py_ssize_t): + * PyObject_Free(info.strides) # <<<<<<<<<<<<<< + * # info.shape was stored after info.strides in the same block + * + */ + PyObject_Free(__pyx_v_info->strides); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":340 + * if PyArray_HASFIELDS(self): + * PyObject_Free(info.format) + * if sizeof(npy_intp) != sizeof(Py_ssize_t): # <<<<<<<<<<<<<< + * PyObject_Free(info.strides) + * # info.shape was stored after info.strides in the same block + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":337 + * f[0] = c'\0' # Terminate format string + * + * def __releasebuffer__(ndarray self, Py_buffer* info): # <<<<<<<<<<<<<< + * if PyArray_HASFIELDS(self): + * PyObject_Free(info.format) + */ + + /* function exit code */ + __Pyx_RefNannyFinishContext(); +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":821 + * ctypedef npy_cdouble complex_t + * + * cdef inline object PyArray_MultiIterNew1(a): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(1, a) + * + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_PyArray_MultiIterNew1(PyObject *__pyx_v_a) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("PyArray_MultiIterNew1", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":822 + * + * cdef inline object PyArray_MultiIterNew1(a): + * return PyArray_MultiIterNew(1, a) # <<<<<<<<<<<<<< + * + * cdef inline object PyArray_MultiIterNew2(a, b): + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = PyArray_MultiIterNew(1, ((void *)__pyx_v_a)); if (unlikely(!__pyx_t_1)) __PYX_ERR(1, 822, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_r = __pyx_t_1; + __pyx_t_1 = 0; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":821 + * ctypedef npy_cdouble complex_t + * + * cdef inline object PyArray_MultiIterNew1(a): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(1, a) + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_AddTraceback("numpy.PyArray_MultiIterNew1", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":824 + * return PyArray_MultiIterNew(1, a) + * + * cdef inline object PyArray_MultiIterNew2(a, b): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(2, a, b) + * + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_PyArray_MultiIterNew2(PyObject *__pyx_v_a, PyObject *__pyx_v_b) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("PyArray_MultiIterNew2", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":825 + * + * cdef inline object PyArray_MultiIterNew2(a, b): + * return PyArray_MultiIterNew(2, a, b) # <<<<<<<<<<<<<< + * + * cdef inline object PyArray_MultiIterNew3(a, b, c): + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = PyArray_MultiIterNew(2, ((void *)__pyx_v_a), ((void *)__pyx_v_b)); if (unlikely(!__pyx_t_1)) __PYX_ERR(1, 825, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_r = __pyx_t_1; + __pyx_t_1 = 0; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":824 + * return PyArray_MultiIterNew(1, a) + * + * cdef inline object PyArray_MultiIterNew2(a, b): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(2, a, b) + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_AddTraceback("numpy.PyArray_MultiIterNew2", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":827 + * return PyArray_MultiIterNew(2, a, b) + * + * cdef inline object PyArray_MultiIterNew3(a, b, c): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(3, a, b, c) + * + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_PyArray_MultiIterNew3(PyObject *__pyx_v_a, PyObject *__pyx_v_b, PyObject *__pyx_v_c) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("PyArray_MultiIterNew3", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":828 + * + * cdef inline object PyArray_MultiIterNew3(a, b, c): + * return PyArray_MultiIterNew(3, a, b, c) # <<<<<<<<<<<<<< + * + * cdef inline object PyArray_MultiIterNew4(a, b, c, d): + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = PyArray_MultiIterNew(3, ((void *)__pyx_v_a), ((void *)__pyx_v_b), ((void *)__pyx_v_c)); if (unlikely(!__pyx_t_1)) __PYX_ERR(1, 828, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_r = __pyx_t_1; + __pyx_t_1 = 0; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":827 + * return PyArray_MultiIterNew(2, a, b) + * + * cdef inline object PyArray_MultiIterNew3(a, b, c): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(3, a, b, c) + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_AddTraceback("numpy.PyArray_MultiIterNew3", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":830 + * return PyArray_MultiIterNew(3, a, b, c) + * + * cdef inline object PyArray_MultiIterNew4(a, b, c, d): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(4, a, b, c, d) + * + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_PyArray_MultiIterNew4(PyObject *__pyx_v_a, PyObject *__pyx_v_b, PyObject *__pyx_v_c, PyObject *__pyx_v_d) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("PyArray_MultiIterNew4", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":831 + * + * cdef inline object PyArray_MultiIterNew4(a, b, c, d): + * return PyArray_MultiIterNew(4, a, b, c, d) # <<<<<<<<<<<<<< + * + * cdef inline object PyArray_MultiIterNew5(a, b, c, d, e): + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = PyArray_MultiIterNew(4, ((void *)__pyx_v_a), ((void *)__pyx_v_b), ((void *)__pyx_v_c), ((void *)__pyx_v_d)); if (unlikely(!__pyx_t_1)) __PYX_ERR(1, 831, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_r = __pyx_t_1; + __pyx_t_1 = 0; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":830 + * return PyArray_MultiIterNew(3, a, b, c) + * + * cdef inline object PyArray_MultiIterNew4(a, b, c, d): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(4, a, b, c, d) + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_AddTraceback("numpy.PyArray_MultiIterNew4", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":833 + * return PyArray_MultiIterNew(4, a, b, c, d) + * + * cdef inline object PyArray_MultiIterNew5(a, b, c, d, e): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(5, a, b, c, d, e) + * + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_PyArray_MultiIterNew5(PyObject *__pyx_v_a, PyObject *__pyx_v_b, PyObject *__pyx_v_c, PyObject *__pyx_v_d, PyObject *__pyx_v_e) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("PyArray_MultiIterNew5", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":834 + * + * cdef inline object PyArray_MultiIterNew5(a, b, c, d, e): + * return PyArray_MultiIterNew(5, a, b, c, d, e) # <<<<<<<<<<<<<< + * + * cdef inline tuple PyDataType_SHAPE(dtype d): + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = PyArray_MultiIterNew(5, ((void *)__pyx_v_a), ((void *)__pyx_v_b), ((void *)__pyx_v_c), ((void *)__pyx_v_d), ((void *)__pyx_v_e)); if (unlikely(!__pyx_t_1)) __PYX_ERR(1, 834, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_r = __pyx_t_1; + __pyx_t_1 = 0; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":833 + * return PyArray_MultiIterNew(4, a, b, c, d) + * + * cdef inline object PyArray_MultiIterNew5(a, b, c, d, e): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(5, a, b, c, d, e) + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_AddTraceback("numpy.PyArray_MultiIterNew5", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":836 + * return PyArray_MultiIterNew(5, a, b, c, d, e) + * + * cdef inline tuple PyDataType_SHAPE(dtype d): # <<<<<<<<<<<<<< + * if PyDataType_HASSUBARRAY(d): + * return d.subarray.shape + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_PyDataType_SHAPE(PyArray_Descr *__pyx_v_d) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + int __pyx_t_1; + __Pyx_RefNannySetupContext("PyDataType_SHAPE", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":837 + * + * cdef inline tuple PyDataType_SHAPE(dtype d): + * if PyDataType_HASSUBARRAY(d): # <<<<<<<<<<<<<< + * return d.subarray.shape + * else: + */ + __pyx_t_1 = (PyDataType_HASSUBARRAY(__pyx_v_d) != 0); + if (__pyx_t_1) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":838 + * cdef inline tuple PyDataType_SHAPE(dtype d): + * if PyDataType_HASSUBARRAY(d): + * return d.subarray.shape # <<<<<<<<<<<<<< + * else: + * return () + */ + __Pyx_XDECREF(__pyx_r); + __Pyx_INCREF(((PyObject*)__pyx_v_d->subarray->shape)); + __pyx_r = ((PyObject*)__pyx_v_d->subarray->shape); + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":837 + * + * cdef inline tuple PyDataType_SHAPE(dtype d): + * if PyDataType_HASSUBARRAY(d): # <<<<<<<<<<<<<< + * return d.subarray.shape + * else: + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":840 + * return d.subarray.shape + * else: + * return () # <<<<<<<<<<<<<< + * + * cdef inline char* _util_dtypestring(dtype descr, char* f, char* end, int* offset) except NULL: + */ + /*else*/ { + __Pyx_XDECREF(__pyx_r); + __Pyx_INCREF(__pyx_empty_tuple); + __pyx_r = __pyx_empty_tuple; + goto __pyx_L0; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":836 + * return PyArray_MultiIterNew(5, a, b, c, d, e) + * + * cdef inline tuple PyDataType_SHAPE(dtype d): # <<<<<<<<<<<<<< + * if PyDataType_HASSUBARRAY(d): + * return d.subarray.shape + */ + + /* function exit code */ + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":842 + * return () + * + * cdef inline char* _util_dtypestring(dtype descr, char* f, char* end, int* offset) except NULL: # <<<<<<<<<<<<<< + * # Recursive utility function used in __getbuffer__ to get format + * # string. The new location in the format string is returned. + */ + +static CYTHON_INLINE char *__pyx_f_5numpy__util_dtypestring(PyArray_Descr *__pyx_v_descr, char *__pyx_v_f, char *__pyx_v_end, int *__pyx_v_offset) { + PyArray_Descr *__pyx_v_child = 0; + int __pyx_v_endian_detector; + int __pyx_v_little_endian; + PyObject *__pyx_v_fields = 0; + PyObject *__pyx_v_childname = NULL; + PyObject *__pyx_v_new_offset = NULL; + PyObject *__pyx_v_t = NULL; + char *__pyx_r; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + Py_ssize_t __pyx_t_2; + PyObject *__pyx_t_3 = NULL; + PyObject *__pyx_t_4 = NULL; + int __pyx_t_5; + int __pyx_t_6; + int __pyx_t_7; + long __pyx_t_8; + char *__pyx_t_9; + __Pyx_RefNannySetupContext("_util_dtypestring", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":847 + * + * cdef dtype child + * cdef int endian_detector = 1 # <<<<<<<<<<<<<< + * cdef bint little_endian = ((&endian_detector)[0] != 0) + * cdef tuple fields + */ + __pyx_v_endian_detector = 1; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":848 + * cdef dtype child + * cdef int endian_detector = 1 + * cdef bint little_endian = ((&endian_detector)[0] != 0) # <<<<<<<<<<<<<< + * cdef tuple fields + * + */ + __pyx_v_little_endian = ((((char *)(&__pyx_v_endian_detector))[0]) != 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":851 + * cdef tuple fields + * + * for childname in descr.names: # <<<<<<<<<<<<<< + * fields = descr.fields[childname] + * child, new_offset = fields + */ + if (unlikely(__pyx_v_descr->names == Py_None)) { + PyErr_SetString(PyExc_TypeError, "'NoneType' object is not iterable"); + __PYX_ERR(1, 851, __pyx_L1_error) + } + __pyx_t_1 = __pyx_v_descr->names; __Pyx_INCREF(__pyx_t_1); __pyx_t_2 = 0; + for (;;) { + if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_1)) break; + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_3 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_3); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(1, 851, __pyx_L1_error) + #else + __pyx_t_3 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 851, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + #endif + __Pyx_XDECREF_SET(__pyx_v_childname, __pyx_t_3); + __pyx_t_3 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":852 + * + * for childname in descr.names: + * fields = descr.fields[childname] # <<<<<<<<<<<<<< + * child, new_offset = fields + * + */ + if (unlikely(__pyx_v_descr->fields == Py_None)) { + PyErr_SetString(PyExc_TypeError, "'NoneType' object is not subscriptable"); + __PYX_ERR(1, 852, __pyx_L1_error) + } + __pyx_t_3 = __Pyx_PyDict_GetItem(__pyx_v_descr->fields, __pyx_v_childname); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 852, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + if (!(likely(PyTuple_CheckExact(__pyx_t_3))||((__pyx_t_3) == Py_None)||(PyErr_Format(PyExc_TypeError, "Expected %.16s, got %.200s", "tuple", Py_TYPE(__pyx_t_3)->tp_name), 0))) __PYX_ERR(1, 852, __pyx_L1_error) + __Pyx_XDECREF_SET(__pyx_v_fields, ((PyObject*)__pyx_t_3)); + __pyx_t_3 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":853 + * for childname in descr.names: + * fields = descr.fields[childname] + * child, new_offset = fields # <<<<<<<<<<<<<< + * + * if (end - f) - (new_offset - offset[0]) < 15: + */ + if (likely(__pyx_v_fields != Py_None)) { + PyObject* sequence = __pyx_v_fields; + Py_ssize_t size = __Pyx_PySequence_SIZE(sequence); + if (unlikely(size != 2)) { + if (size > 2) __Pyx_RaiseTooManyValuesError(2); + else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); + __PYX_ERR(1, 853, __pyx_L1_error) + } + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_3 = PyTuple_GET_ITEM(sequence, 0); + __pyx_t_4 = PyTuple_GET_ITEM(sequence, 1); + __Pyx_INCREF(__pyx_t_3); + __Pyx_INCREF(__pyx_t_4); + #else + __pyx_t_3 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 853, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 853, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + #endif + } else { + __Pyx_RaiseNoneNotIterableError(); __PYX_ERR(1, 853, __pyx_L1_error) + } + if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_dtype))))) __PYX_ERR(1, 853, __pyx_L1_error) + __Pyx_XDECREF_SET(__pyx_v_child, ((PyArray_Descr *)__pyx_t_3)); + __pyx_t_3 = 0; + __Pyx_XDECREF_SET(__pyx_v_new_offset, __pyx_t_4); + __pyx_t_4 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":855 + * child, new_offset = fields + * + * if (end - f) - (new_offset - offset[0]) < 15: # <<<<<<<<<<<<<< + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") + * + */ + __pyx_t_4 = __Pyx_PyInt_From_int((__pyx_v_offset[0])); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 855, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyNumber_Subtract(__pyx_v_new_offset, __pyx_t_4); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 855, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_5 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_5 == (int)-1) && PyErr_Occurred())) __PYX_ERR(1, 855, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = ((((__pyx_v_end - __pyx_v_f) - ((int)__pyx_t_5)) < 15) != 0); + if (unlikely(__pyx_t_6)) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":856 + * + * if (end - f) - (new_offset - offset[0]) < 15: + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") # <<<<<<<<<<<<<< + * + * if ((child.byteorder == c'>' and little_endian) or + */ + __pyx_t_3 = __Pyx_PyObject_Call(__pyx_builtin_RuntimeError, __pyx_tuple__4, NULL); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 856, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_Raise(__pyx_t_3, 0, 0, 0); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __PYX_ERR(1, 856, __pyx_L1_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":855 + * child, new_offset = fields + * + * if (end - f) - (new_offset - offset[0]) < 15: # <<<<<<<<<<<<<< + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") + * + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":858 + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") + * + * if ((child.byteorder == c'>' and little_endian) or # <<<<<<<<<<<<<< + * (child.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + */ + __pyx_t_7 = ((__pyx_v_child->byteorder == '>') != 0); + if (!__pyx_t_7) { + goto __pyx_L8_next_or; + } else { + } + __pyx_t_7 = (__pyx_v_little_endian != 0); + if (!__pyx_t_7) { + } else { + __pyx_t_6 = __pyx_t_7; + goto __pyx_L7_bool_binop_done; + } + __pyx_L8_next_or:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":859 + * + * if ((child.byteorder == c'>' and little_endian) or + * (child.byteorder == c'<' and not little_endian)): # <<<<<<<<<<<<<< + * raise ValueError(u"Non-native byte order not supported") + * # One could encode it in the format string and have Cython + */ + __pyx_t_7 = ((__pyx_v_child->byteorder == '<') != 0); + if (__pyx_t_7) { + } else { + __pyx_t_6 = __pyx_t_7; + goto __pyx_L7_bool_binop_done; + } + __pyx_t_7 = ((!(__pyx_v_little_endian != 0)) != 0); + __pyx_t_6 = __pyx_t_7; + __pyx_L7_bool_binop_done:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":858 + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") + * + * if ((child.byteorder == c'>' and little_endian) or # <<<<<<<<<<<<<< + * (child.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + */ + if (unlikely(__pyx_t_6)) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":860 + * if ((child.byteorder == c'>' and little_endian) or + * (child.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") # <<<<<<<<<<<<<< + * # One could encode it in the format string and have Cython + * # complain instead, BUT: < and > in format strings also imply + */ + __pyx_t_3 = __Pyx_PyObject_Call(__pyx_builtin_ValueError, __pyx_tuple__3, NULL); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 860, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_Raise(__pyx_t_3, 0, 0, 0); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __PYX_ERR(1, 860, __pyx_L1_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":858 + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") + * + * if ((child.byteorder == c'>' and little_endian) or # <<<<<<<<<<<<<< + * (child.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":870 + * + * # Output padding bytes + * while offset[0] < new_offset: # <<<<<<<<<<<<<< + * f[0] = 120 # "x"; pad byte + * f += 1 + */ + while (1) { + __pyx_t_3 = __Pyx_PyInt_From_int((__pyx_v_offset[0])); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 870, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_t_3, __pyx_v_new_offset, Py_LT); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 870, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 870, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (!__pyx_t_6) break; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":871 + * # Output padding bytes + * while offset[0] < new_offset: + * f[0] = 120 # "x"; pad byte # <<<<<<<<<<<<<< + * f += 1 + * offset[0] += 1 + */ + (__pyx_v_f[0]) = 0x78; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":872 + * while offset[0] < new_offset: + * f[0] = 120 # "x"; pad byte + * f += 1 # <<<<<<<<<<<<<< + * offset[0] += 1 + * + */ + __pyx_v_f = (__pyx_v_f + 1); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":873 + * f[0] = 120 # "x"; pad byte + * f += 1 + * offset[0] += 1 # <<<<<<<<<<<<<< + * + * offset[0] += child.itemsize + */ + __pyx_t_8 = 0; + (__pyx_v_offset[__pyx_t_8]) = ((__pyx_v_offset[__pyx_t_8]) + 1); + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":875 + * offset[0] += 1 + * + * offset[0] += child.itemsize # <<<<<<<<<<<<<< + * + * if not PyDataType_HASFIELDS(child): + */ + __pyx_t_8 = 0; + (__pyx_v_offset[__pyx_t_8]) = ((__pyx_v_offset[__pyx_t_8]) + __pyx_v_child->elsize); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":877 + * offset[0] += child.itemsize + * + * if not PyDataType_HASFIELDS(child): # <<<<<<<<<<<<<< + * t = child.type_num + * if end - f < 5: + */ + __pyx_t_6 = ((!(PyDataType_HASFIELDS(__pyx_v_child) != 0)) != 0); + if (__pyx_t_6) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":878 + * + * if not PyDataType_HASFIELDS(child): + * t = child.type_num # <<<<<<<<<<<<<< + * if end - f < 5: + * raise RuntimeError(u"Format string allocated too short.") + */ + __pyx_t_4 = __Pyx_PyInt_From_int(__pyx_v_child->type_num); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 878, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __Pyx_XDECREF_SET(__pyx_v_t, __pyx_t_4); + __pyx_t_4 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":879 + * if not PyDataType_HASFIELDS(child): + * t = child.type_num + * if end - f < 5: # <<<<<<<<<<<<<< + * raise RuntimeError(u"Format string allocated too short.") + * + */ + __pyx_t_6 = (((__pyx_v_end - __pyx_v_f) < 5) != 0); + if (unlikely(__pyx_t_6)) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":880 + * t = child.type_num + * if end - f < 5: + * raise RuntimeError(u"Format string allocated too short.") # <<<<<<<<<<<<<< + * + * # Until ticket #99 is fixed, use integers to avoid warnings + */ + __pyx_t_4 = __Pyx_PyObject_Call(__pyx_builtin_RuntimeError, __pyx_tuple__5, NULL); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 880, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __Pyx_Raise(__pyx_t_4, 0, 0, 0); + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __PYX_ERR(1, 880, __pyx_L1_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":879 + * if not PyDataType_HASFIELDS(child): + * t = child.type_num + * if end - f < 5: # <<<<<<<<<<<<<< + * raise RuntimeError(u"Format string allocated too short.") + * + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":883 + * + * # Until ticket #99 is fixed, use integers to avoid warnings + * if t == NPY_BYTE: f[0] = 98 #"b" # <<<<<<<<<<<<<< + * elif t == NPY_UBYTE: f[0] = 66 #"B" + * elif t == NPY_SHORT: f[0] = 104 #"h" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_BYTE); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 883, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 883, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 883, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 98; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":884 + * # Until ticket #99 is fixed, use integers to avoid warnings + * if t == NPY_BYTE: f[0] = 98 #"b" + * elif t == NPY_UBYTE: f[0] = 66 #"B" # <<<<<<<<<<<<<< + * elif t == NPY_SHORT: f[0] = 104 #"h" + * elif t == NPY_USHORT: f[0] = 72 #"H" + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_UBYTE); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 884, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 884, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 884, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 66; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":885 + * if t == NPY_BYTE: f[0] = 98 #"b" + * elif t == NPY_UBYTE: f[0] = 66 #"B" + * elif t == NPY_SHORT: f[0] = 104 #"h" # <<<<<<<<<<<<<< + * elif t == NPY_USHORT: f[0] = 72 #"H" + * elif t == NPY_INT: f[0] = 105 #"i" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_SHORT); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 885, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 885, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 885, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x68; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":886 + * elif t == NPY_UBYTE: f[0] = 66 #"B" + * elif t == NPY_SHORT: f[0] = 104 #"h" + * elif t == NPY_USHORT: f[0] = 72 #"H" # <<<<<<<<<<<<<< + * elif t == NPY_INT: f[0] = 105 #"i" + * elif t == NPY_UINT: f[0] = 73 #"I" + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_USHORT); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 886, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 886, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 886, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 72; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":887 + * elif t == NPY_SHORT: f[0] = 104 #"h" + * elif t == NPY_USHORT: f[0] = 72 #"H" + * elif t == NPY_INT: f[0] = 105 #"i" # <<<<<<<<<<<<<< + * elif t == NPY_UINT: f[0] = 73 #"I" + * elif t == NPY_LONG: f[0] = 108 #"l" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_INT); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 887, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 887, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 887, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x69; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":888 + * elif t == NPY_USHORT: f[0] = 72 #"H" + * elif t == NPY_INT: f[0] = 105 #"i" + * elif t == NPY_UINT: f[0] = 73 #"I" # <<<<<<<<<<<<<< + * elif t == NPY_LONG: f[0] = 108 #"l" + * elif t == NPY_ULONG: f[0] = 76 #"L" + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_UINT); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 888, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 888, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 888, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 73; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":889 + * elif t == NPY_INT: f[0] = 105 #"i" + * elif t == NPY_UINT: f[0] = 73 #"I" + * elif t == NPY_LONG: f[0] = 108 #"l" # <<<<<<<<<<<<<< + * elif t == NPY_ULONG: f[0] = 76 #"L" + * elif t == NPY_LONGLONG: f[0] = 113 #"q" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_LONG); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 889, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 889, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 889, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x6C; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":890 + * elif t == NPY_UINT: f[0] = 73 #"I" + * elif t == NPY_LONG: f[0] = 108 #"l" + * elif t == NPY_ULONG: f[0] = 76 #"L" # <<<<<<<<<<<<<< + * elif t == NPY_LONGLONG: f[0] = 113 #"q" + * elif t == NPY_ULONGLONG: f[0] = 81 #"Q" + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_ULONG); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 890, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 890, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 890, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 76; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":891 + * elif t == NPY_LONG: f[0] = 108 #"l" + * elif t == NPY_ULONG: f[0] = 76 #"L" + * elif t == NPY_LONGLONG: f[0] = 113 #"q" # <<<<<<<<<<<<<< + * elif t == NPY_ULONGLONG: f[0] = 81 #"Q" + * elif t == NPY_FLOAT: f[0] = 102 #"f" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_LONGLONG); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 891, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 891, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 891, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x71; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":892 + * elif t == NPY_ULONG: f[0] = 76 #"L" + * elif t == NPY_LONGLONG: f[0] = 113 #"q" + * elif t == NPY_ULONGLONG: f[0] = 81 #"Q" # <<<<<<<<<<<<<< + * elif t == NPY_FLOAT: f[0] = 102 #"f" + * elif t == NPY_DOUBLE: f[0] = 100 #"d" + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_ULONGLONG); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 892, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 892, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 892, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 81; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":893 + * elif t == NPY_LONGLONG: f[0] = 113 #"q" + * elif t == NPY_ULONGLONG: f[0] = 81 #"Q" + * elif t == NPY_FLOAT: f[0] = 102 #"f" # <<<<<<<<<<<<<< + * elif t == NPY_DOUBLE: f[0] = 100 #"d" + * elif t == NPY_LONGDOUBLE: f[0] = 103 #"g" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_FLOAT); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 893, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 893, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 893, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x66; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":894 + * elif t == NPY_ULONGLONG: f[0] = 81 #"Q" + * elif t == NPY_FLOAT: f[0] = 102 #"f" + * elif t == NPY_DOUBLE: f[0] = 100 #"d" # <<<<<<<<<<<<<< + * elif t == NPY_LONGDOUBLE: f[0] = 103 #"g" + * elif t == NPY_CFLOAT: f[0] = 90; f[1] = 102; f += 1 # Zf + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_DOUBLE); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 894, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 894, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 894, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x64; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":895 + * elif t == NPY_FLOAT: f[0] = 102 #"f" + * elif t == NPY_DOUBLE: f[0] = 100 #"d" + * elif t == NPY_LONGDOUBLE: f[0] = 103 #"g" # <<<<<<<<<<<<<< + * elif t == NPY_CFLOAT: f[0] = 90; f[1] = 102; f += 1 # Zf + * elif t == NPY_CDOUBLE: f[0] = 90; f[1] = 100; f += 1 # Zd + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_LONGDOUBLE); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 895, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 895, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 895, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x67; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":896 + * elif t == NPY_DOUBLE: f[0] = 100 #"d" + * elif t == NPY_LONGDOUBLE: f[0] = 103 #"g" + * elif t == NPY_CFLOAT: f[0] = 90; f[1] = 102; f += 1 # Zf # <<<<<<<<<<<<<< + * elif t == NPY_CDOUBLE: f[0] = 90; f[1] = 100; f += 1 # Zd + * elif t == NPY_CLONGDOUBLE: f[0] = 90; f[1] = 103; f += 1 # Zg + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_CFLOAT); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 896, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 896, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 896, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 90; + (__pyx_v_f[1]) = 0x66; + __pyx_v_f = (__pyx_v_f + 1); + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":897 + * elif t == NPY_LONGDOUBLE: f[0] = 103 #"g" + * elif t == NPY_CFLOAT: f[0] = 90; f[1] = 102; f += 1 # Zf + * elif t == NPY_CDOUBLE: f[0] = 90; f[1] = 100; f += 1 # Zd # <<<<<<<<<<<<<< + * elif t == NPY_CLONGDOUBLE: f[0] = 90; f[1] = 103; f += 1 # Zg + * elif t == NPY_OBJECT: f[0] = 79 #"O" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_CDOUBLE); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 897, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 897, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 897, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 90; + (__pyx_v_f[1]) = 0x64; + __pyx_v_f = (__pyx_v_f + 1); + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":898 + * elif t == NPY_CFLOAT: f[0] = 90; f[1] = 102; f += 1 # Zf + * elif t == NPY_CDOUBLE: f[0] = 90; f[1] = 100; f += 1 # Zd + * elif t == NPY_CLONGDOUBLE: f[0] = 90; f[1] = 103; f += 1 # Zg # <<<<<<<<<<<<<< + * elif t == NPY_OBJECT: f[0] = 79 #"O" + * else: + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_CLONGDOUBLE); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 898, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 898, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 898, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 90; + (__pyx_v_f[1]) = 0x67; + __pyx_v_f = (__pyx_v_f + 1); + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":899 + * elif t == NPY_CDOUBLE: f[0] = 90; f[1] = 100; f += 1 # Zd + * elif t == NPY_CLONGDOUBLE: f[0] = 90; f[1] = 103; f += 1 # Zg + * elif t == NPY_OBJECT: f[0] = 79 #"O" # <<<<<<<<<<<<<< + * else: + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_OBJECT); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 899, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 899, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 899, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (likely(__pyx_t_6)) { + (__pyx_v_f[0]) = 79; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":901 + * elif t == NPY_OBJECT: f[0] = 79 #"O" + * else: + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) # <<<<<<<<<<<<<< + * f += 1 + * else: + */ + /*else*/ { + __pyx_t_3 = __Pyx_PyUnicode_FormatSafe(__pyx_kp_u_unknown_dtype_code_in_numpy_pxd, __pyx_v_t); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 901, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = __Pyx_PyObject_CallOneArg(__pyx_builtin_ValueError, __pyx_t_3); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 901, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __Pyx_Raise(__pyx_t_4, 0, 0, 0); + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __PYX_ERR(1, 901, __pyx_L1_error) + } + __pyx_L15:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":902 + * else: + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) + * f += 1 # <<<<<<<<<<<<<< + * else: + * # Cython ignores struct boundary information ("T{...}"), + */ + __pyx_v_f = (__pyx_v_f + 1); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":877 + * offset[0] += child.itemsize + * + * if not PyDataType_HASFIELDS(child): # <<<<<<<<<<<<<< + * t = child.type_num + * if end - f < 5: + */ + goto __pyx_L13; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":906 + * # Cython ignores struct boundary information ("T{...}"), + * # so don't output it + * f = _util_dtypestring(child, f, end, offset) # <<<<<<<<<<<<<< + * return f + * + */ + /*else*/ { + __pyx_t_9 = __pyx_f_5numpy__util_dtypestring(__pyx_v_child, __pyx_v_f, __pyx_v_end, __pyx_v_offset); if (unlikely(__pyx_t_9 == ((char *)NULL))) __PYX_ERR(1, 906, __pyx_L1_error) + __pyx_v_f = __pyx_t_9; + } + __pyx_L13:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":851 + * cdef tuple fields + * + * for childname in descr.names: # <<<<<<<<<<<<<< + * fields = descr.fields[childname] + * child, new_offset = fields + */ + } + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":907 + * # so don't output it + * f = _util_dtypestring(child, f, end, offset) + * return f # <<<<<<<<<<<<<< + * + * + */ + __pyx_r = __pyx_v_f; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":842 + * return () + * + * cdef inline char* _util_dtypestring(dtype descr, char* f, char* end, int* offset) except NULL: # <<<<<<<<<<<<<< + * # Recursive utility function used in __getbuffer__ to get format + * # string. The new location in the format string is returned. + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_XDECREF(__pyx_t_3); + __Pyx_XDECREF(__pyx_t_4); + __Pyx_AddTraceback("numpy._util_dtypestring", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = NULL; + __pyx_L0:; + __Pyx_XDECREF((PyObject *)__pyx_v_child); + __Pyx_XDECREF(__pyx_v_fields); + __Pyx_XDECREF(__pyx_v_childname); + __Pyx_XDECREF(__pyx_v_new_offset); + __Pyx_XDECREF(__pyx_v_t); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1022 + * int _import_umath() except -1 + * + * cdef inline void set_array_base(ndarray arr, object base): # <<<<<<<<<<<<<< + * Py_INCREF(base) # important to do this before stealing the reference below! + * PyArray_SetBaseObject(arr, base) + */ + +static CYTHON_INLINE void __pyx_f_5numpy_set_array_base(PyArrayObject *__pyx_v_arr, PyObject *__pyx_v_base) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("set_array_base", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1023 + * + * cdef inline void set_array_base(ndarray arr, object base): + * Py_INCREF(base) # important to do this before stealing the reference below! # <<<<<<<<<<<<<< + * PyArray_SetBaseObject(arr, base) + * + */ + Py_INCREF(__pyx_v_base); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1024 + * cdef inline void set_array_base(ndarray arr, object base): + * Py_INCREF(base) # important to do this before stealing the reference below! + * PyArray_SetBaseObject(arr, base) # <<<<<<<<<<<<<< + * + * cdef inline object get_array_base(ndarray arr): + */ + (void)(PyArray_SetBaseObject(__pyx_v_arr, __pyx_v_base)); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1022 + * int _import_umath() except -1 + * + * cdef inline void set_array_base(ndarray arr, object base): # <<<<<<<<<<<<<< + * Py_INCREF(base) # important to do this before stealing the reference below! + * PyArray_SetBaseObject(arr, base) + */ + + /* function exit code */ + __Pyx_RefNannyFinishContext(); +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1026 + * PyArray_SetBaseObject(arr, base) + * + * cdef inline object get_array_base(ndarray arr): # <<<<<<<<<<<<<< + * base = PyArray_BASE(arr) + * if base is NULL: + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_get_array_base(PyArrayObject *__pyx_v_arr) { + PyObject *__pyx_v_base; + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + int __pyx_t_1; + __Pyx_RefNannySetupContext("get_array_base", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1027 + * + * cdef inline object get_array_base(ndarray arr): + * base = PyArray_BASE(arr) # <<<<<<<<<<<<<< + * if base is NULL: + * return None + */ + __pyx_v_base = PyArray_BASE(__pyx_v_arr); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1028 + * cdef inline object get_array_base(ndarray arr): + * base = PyArray_BASE(arr) + * if base is NULL: # <<<<<<<<<<<<<< + * return None + * return base + */ + __pyx_t_1 = ((__pyx_v_base == NULL) != 0); + if (__pyx_t_1) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1029 + * base = PyArray_BASE(arr) + * if base is NULL: + * return None # <<<<<<<<<<<<<< + * return base + * + */ + __Pyx_XDECREF(__pyx_r); + __pyx_r = Py_None; __Pyx_INCREF(Py_None); + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1028 + * cdef inline object get_array_base(ndarray arr): + * base = PyArray_BASE(arr) + * if base is NULL: # <<<<<<<<<<<<<< + * return None + * return base + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1030 + * if base is NULL: + * return None + * return base # <<<<<<<<<<<<<< + * + * # Versions of the import_* functions which are more suitable for + */ + __Pyx_XDECREF(__pyx_r); + __Pyx_INCREF(((PyObject *)__pyx_v_base)); + __pyx_r = ((PyObject *)__pyx_v_base); + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1026 + * PyArray_SetBaseObject(arr, base) + * + * cdef inline object get_array_base(ndarray arr): # <<<<<<<<<<<<<< + * base = PyArray_BASE(arr) + * if base is NULL: + */ + + /* function exit code */ + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1034 + * # Versions of the import_* functions which are more suitable for + * # Cython code. + * cdef inline int import_array() except -1: # <<<<<<<<<<<<<< + * try: + * _import_array() + */ + +static CYTHON_INLINE int __pyx_f_5numpy_import_array(void) { + int __pyx_r; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + PyObject *__pyx_t_2 = NULL; + PyObject *__pyx_t_3 = NULL; + int __pyx_t_4; + PyObject *__pyx_t_5 = NULL; + PyObject *__pyx_t_6 = NULL; + PyObject *__pyx_t_7 = NULL; + PyObject *__pyx_t_8 = NULL; + __Pyx_RefNannySetupContext("import_array", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1035 + * # Cython code. + * cdef inline int import_array() except -1: + * try: # <<<<<<<<<<<<<< + * _import_array() + * except Exception: + */ + { + __Pyx_PyThreadState_declare + __Pyx_PyThreadState_assign + __Pyx_ExceptionSave(&__pyx_t_1, &__pyx_t_2, &__pyx_t_3); + __Pyx_XGOTREF(__pyx_t_1); + __Pyx_XGOTREF(__pyx_t_2); + __Pyx_XGOTREF(__pyx_t_3); + /*try:*/ { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1036 + * cdef inline int import_array() except -1: + * try: + * _import_array() # <<<<<<<<<<<<<< + * except Exception: + * raise ImportError("numpy.core.multiarray failed to import") + */ + __pyx_t_4 = _import_array(); if (unlikely(__pyx_t_4 == ((int)-1))) __PYX_ERR(1, 1036, __pyx_L3_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1035 + * # Cython code. + * cdef inline int import_array() except -1: + * try: # <<<<<<<<<<<<<< + * _import_array() + * except Exception: + */ + } + __Pyx_XDECREF(__pyx_t_1); __pyx_t_1 = 0; + __Pyx_XDECREF(__pyx_t_2); __pyx_t_2 = 0; + __Pyx_XDECREF(__pyx_t_3); __pyx_t_3 = 0; + goto __pyx_L8_try_end; + __pyx_L3_error:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1037 + * try: + * _import_array() + * except Exception: # <<<<<<<<<<<<<< + * raise ImportError("numpy.core.multiarray failed to import") + * + */ + __pyx_t_4 = __Pyx_PyErr_ExceptionMatches(((PyObject *)(&((PyTypeObject*)PyExc_Exception)[0]))); + if (__pyx_t_4) { + __Pyx_AddTraceback("numpy.import_array", __pyx_clineno, __pyx_lineno, __pyx_filename); + if (__Pyx_GetException(&__pyx_t_5, &__pyx_t_6, &__pyx_t_7) < 0) __PYX_ERR(1, 1037, __pyx_L5_except_error) + __Pyx_GOTREF(__pyx_t_5); + __Pyx_GOTREF(__pyx_t_6); + __Pyx_GOTREF(__pyx_t_7); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1038 + * _import_array() + * except Exception: + * raise ImportError("numpy.core.multiarray failed to import") # <<<<<<<<<<<<<< + * + * cdef inline int import_umath() except -1: + */ + __pyx_t_8 = __Pyx_PyObject_Call(__pyx_builtin_ImportError, __pyx_tuple__6, NULL); if (unlikely(!__pyx_t_8)) __PYX_ERR(1, 1038, __pyx_L5_except_error) + __Pyx_GOTREF(__pyx_t_8); + __Pyx_Raise(__pyx_t_8, 0, 0, 0); + __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; + __PYX_ERR(1, 1038, __pyx_L5_except_error) + } + goto __pyx_L5_except_error; + __pyx_L5_except_error:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1035 + * # Cython code. + * cdef inline int import_array() except -1: + * try: # <<<<<<<<<<<<<< + * _import_array() + * except Exception: + */ + __Pyx_XGIVEREF(__pyx_t_1); + __Pyx_XGIVEREF(__pyx_t_2); + __Pyx_XGIVEREF(__pyx_t_3); + __Pyx_ExceptionReset(__pyx_t_1, __pyx_t_2, __pyx_t_3); + goto __pyx_L1_error; + __pyx_L8_try_end:; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1034 + * # Versions of the import_* functions which are more suitable for + * # Cython code. + * cdef inline int import_array() except -1: # <<<<<<<<<<<<<< + * try: + * _import_array() + */ + + /* function exit code */ + __pyx_r = 0; + goto __pyx_L0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_5); + __Pyx_XDECREF(__pyx_t_6); + __Pyx_XDECREF(__pyx_t_7); + __Pyx_XDECREF(__pyx_t_8); + __Pyx_AddTraceback("numpy.import_array", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = -1; + __pyx_L0:; + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1040 + * raise ImportError("numpy.core.multiarray failed to import") + * + * cdef inline int import_umath() except -1: # <<<<<<<<<<<<<< + * try: + * _import_umath() + */ + +static CYTHON_INLINE int __pyx_f_5numpy_import_umath(void) { + int __pyx_r; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + PyObject *__pyx_t_2 = NULL; + PyObject *__pyx_t_3 = NULL; + int __pyx_t_4; + PyObject *__pyx_t_5 = NULL; + PyObject *__pyx_t_6 = NULL; + PyObject *__pyx_t_7 = NULL; + PyObject *__pyx_t_8 = NULL; + __Pyx_RefNannySetupContext("import_umath", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1041 + * + * cdef inline int import_umath() except -1: + * try: # <<<<<<<<<<<<<< + * _import_umath() + * except Exception: + */ + { + __Pyx_PyThreadState_declare + __Pyx_PyThreadState_assign + __Pyx_ExceptionSave(&__pyx_t_1, &__pyx_t_2, &__pyx_t_3); + __Pyx_XGOTREF(__pyx_t_1); + __Pyx_XGOTREF(__pyx_t_2); + __Pyx_XGOTREF(__pyx_t_3); + /*try:*/ { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1042 + * cdef inline int import_umath() except -1: + * try: + * _import_umath() # <<<<<<<<<<<<<< + * except Exception: + * raise ImportError("numpy.core.umath failed to import") + */ + __pyx_t_4 = _import_umath(); if (unlikely(__pyx_t_4 == ((int)-1))) __PYX_ERR(1, 1042, __pyx_L3_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1041 + * + * cdef inline int import_umath() except -1: + * try: # <<<<<<<<<<<<<< + * _import_umath() + * except Exception: + */ + } + __Pyx_XDECREF(__pyx_t_1); __pyx_t_1 = 0; + __Pyx_XDECREF(__pyx_t_2); __pyx_t_2 = 0; + __Pyx_XDECREF(__pyx_t_3); __pyx_t_3 = 0; + goto __pyx_L8_try_end; + __pyx_L3_error:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1043 + * try: + * _import_umath() + * except Exception: # <<<<<<<<<<<<<< + * raise ImportError("numpy.core.umath failed to import") + * + */ + __pyx_t_4 = __Pyx_PyErr_ExceptionMatches(((PyObject *)(&((PyTypeObject*)PyExc_Exception)[0]))); + if (__pyx_t_4) { + __Pyx_AddTraceback("numpy.import_umath", __pyx_clineno, __pyx_lineno, __pyx_filename); + if (__Pyx_GetException(&__pyx_t_5, &__pyx_t_6, &__pyx_t_7) < 0) __PYX_ERR(1, 1043, __pyx_L5_except_error) + __Pyx_GOTREF(__pyx_t_5); + __Pyx_GOTREF(__pyx_t_6); + __Pyx_GOTREF(__pyx_t_7); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1044 + * _import_umath() + * except Exception: + * raise ImportError("numpy.core.umath failed to import") # <<<<<<<<<<<<<< + * + * cdef inline int import_ufunc() except -1: + */ + __pyx_t_8 = __Pyx_PyObject_Call(__pyx_builtin_ImportError, __pyx_tuple__7, NULL); if (unlikely(!__pyx_t_8)) __PYX_ERR(1, 1044, __pyx_L5_except_error) + __Pyx_GOTREF(__pyx_t_8); + __Pyx_Raise(__pyx_t_8, 0, 0, 0); + __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; + __PYX_ERR(1, 1044, __pyx_L5_except_error) + } + goto __pyx_L5_except_error; + __pyx_L5_except_error:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1041 + * + * cdef inline int import_umath() except -1: + * try: # <<<<<<<<<<<<<< + * _import_umath() + * except Exception: + */ + __Pyx_XGIVEREF(__pyx_t_1); + __Pyx_XGIVEREF(__pyx_t_2); + __Pyx_XGIVEREF(__pyx_t_3); + __Pyx_ExceptionReset(__pyx_t_1, __pyx_t_2, __pyx_t_3); + goto __pyx_L1_error; + __pyx_L8_try_end:; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1040 + * raise ImportError("numpy.core.multiarray failed to import") + * + * cdef inline int import_umath() except -1: # <<<<<<<<<<<<<< + * try: + * _import_umath() + */ + + /* function exit code */ + __pyx_r = 0; + goto __pyx_L0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_5); + __Pyx_XDECREF(__pyx_t_6); + __Pyx_XDECREF(__pyx_t_7); + __Pyx_XDECREF(__pyx_t_8); + __Pyx_AddTraceback("numpy.import_umath", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = -1; + __pyx_L0:; + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1046 + * raise ImportError("numpy.core.umath failed to import") + * + * cdef inline int import_ufunc() except -1: # <<<<<<<<<<<<<< + * try: + * _import_umath() + */ + +static CYTHON_INLINE int __pyx_f_5numpy_import_ufunc(void) { + int __pyx_r; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + PyObject *__pyx_t_2 = NULL; + PyObject *__pyx_t_3 = NULL; + int __pyx_t_4; + PyObject *__pyx_t_5 = NULL; + PyObject *__pyx_t_6 = NULL; + PyObject *__pyx_t_7 = NULL; + PyObject *__pyx_t_8 = NULL; + __Pyx_RefNannySetupContext("import_ufunc", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1047 + * + * cdef inline int import_ufunc() except -1: + * try: # <<<<<<<<<<<<<< + * _import_umath() + * except Exception: + */ + { + __Pyx_PyThreadState_declare + __Pyx_PyThreadState_assign + __Pyx_ExceptionSave(&__pyx_t_1, &__pyx_t_2, &__pyx_t_3); + __Pyx_XGOTREF(__pyx_t_1); + __Pyx_XGOTREF(__pyx_t_2); + __Pyx_XGOTREF(__pyx_t_3); + /*try:*/ { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1048 + * cdef inline int import_ufunc() except -1: + * try: + * _import_umath() # <<<<<<<<<<<<<< + * except Exception: + * raise ImportError("numpy.core.umath failed to import") + */ + __pyx_t_4 = _import_umath(); if (unlikely(__pyx_t_4 == ((int)-1))) __PYX_ERR(1, 1048, __pyx_L3_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1047 + * + * cdef inline int import_ufunc() except -1: + * try: # <<<<<<<<<<<<<< + * _import_umath() + * except Exception: + */ + } + __Pyx_XDECREF(__pyx_t_1); __pyx_t_1 = 0; + __Pyx_XDECREF(__pyx_t_2); __pyx_t_2 = 0; + __Pyx_XDECREF(__pyx_t_3); __pyx_t_3 = 0; + goto __pyx_L8_try_end; + __pyx_L3_error:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1049 + * try: + * _import_umath() + * except Exception: # <<<<<<<<<<<<<< + * raise ImportError("numpy.core.umath failed to import") + */ + __pyx_t_4 = __Pyx_PyErr_ExceptionMatches(((PyObject *)(&((PyTypeObject*)PyExc_Exception)[0]))); + if (__pyx_t_4) { + __Pyx_AddTraceback("numpy.import_ufunc", __pyx_clineno, __pyx_lineno, __pyx_filename); + if (__Pyx_GetException(&__pyx_t_5, &__pyx_t_6, &__pyx_t_7) < 0) __PYX_ERR(1, 1049, __pyx_L5_except_error) + __Pyx_GOTREF(__pyx_t_5); + __Pyx_GOTREF(__pyx_t_6); + __Pyx_GOTREF(__pyx_t_7); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1050 + * _import_umath() + * except Exception: + * raise ImportError("numpy.core.umath failed to import") # <<<<<<<<<<<<<< + */ + __pyx_t_8 = __Pyx_PyObject_Call(__pyx_builtin_ImportError, __pyx_tuple__7, NULL); if (unlikely(!__pyx_t_8)) __PYX_ERR(1, 1050, __pyx_L5_except_error) + __Pyx_GOTREF(__pyx_t_8); + __Pyx_Raise(__pyx_t_8, 0, 0, 0); + __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; + __PYX_ERR(1, 1050, __pyx_L5_except_error) + } + goto __pyx_L5_except_error; + __pyx_L5_except_error:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1047 + * + * cdef inline int import_ufunc() except -1: + * try: # <<<<<<<<<<<<<< + * _import_umath() + * except Exception: + */ + __Pyx_XGIVEREF(__pyx_t_1); + __Pyx_XGIVEREF(__pyx_t_2); + __Pyx_XGIVEREF(__pyx_t_3); + __Pyx_ExceptionReset(__pyx_t_1, __pyx_t_2, __pyx_t_3); + goto __pyx_L1_error; + __pyx_L8_try_end:; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1046 + * raise ImportError("numpy.core.umath failed to import") + * + * cdef inline int import_ufunc() except -1: # <<<<<<<<<<<<<< + * try: + * _import_umath() + */ + + /* function exit code */ + __pyx_r = 0; + goto __pyx_L0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_5); + __Pyx_XDECREF(__pyx_t_6); + __Pyx_XDECREF(__pyx_t_7); + __Pyx_XDECREF(__pyx_t_8); + __Pyx_AddTraceback("numpy.import_ufunc", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = -1; + __pyx_L0:; + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +static PyMethodDef __pyx_methods[] = { + {0, 0, 0, 0} +}; + +#if PY_MAJOR_VERSION >= 3 +#if CYTHON_PEP489_MULTI_PHASE_INIT +static PyObject* __pyx_pymod_create(PyObject *spec, PyModuleDef *def); /*proto*/ +static int __pyx_pymod_exec_average_inner(PyObject* module); /*proto*/ +static PyModuleDef_Slot __pyx_moduledef_slots[] = { + {Py_mod_create, (void*)__pyx_pymod_create}, + {Py_mod_exec, (void*)__pyx_pymod_exec_average_inner}, + {0, NULL} +}; +#endif + +static struct PyModuleDef __pyx_moduledef = { + PyModuleDef_HEAD_INIT, + "average_inner", + __pyx_k_Optimized_cython_functions_for_c, /* m_doc */ + #if CYTHON_PEP489_MULTI_PHASE_INIT + 0, /* m_size */ + #else + -1, /* m_size */ + #endif + __pyx_methods /* m_methods */, + #if CYTHON_PEP489_MULTI_PHASE_INIT + __pyx_moduledef_slots, /* m_slots */ + #else + NULL, /* m_reload */ + #endif + NULL, /* m_traverse */ + NULL, /* m_clear */ + NULL /* m_free */ +}; +#endif +#ifndef CYTHON_SMALL_CODE +#if defined(__clang__) + #define CYTHON_SMALL_CODE +#elif defined(__GNUC__) && (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3)) + #define CYTHON_SMALL_CODE __attribute__((cold)) +#else + #define CYTHON_SMALL_CODE +#endif +#endif + +static __Pyx_StringTabEntry __pyx_string_tab[] = { + {&__pyx_n_s_FAST_VERSION, __pyx_k_FAST_VERSION, sizeof(__pyx_k_FAST_VERSION), 0, 0, 1, 1}, + {&__pyx_kp_u_Format_string_allocated_too_shor, __pyx_k_Format_string_allocated_too_shor, sizeof(__pyx_k_Format_string_allocated_too_shor), 0, 1, 0, 0}, + {&__pyx_kp_u_Format_string_allocated_too_shor_2, __pyx_k_Format_string_allocated_too_shor_2, sizeof(__pyx_k_Format_string_allocated_too_shor_2), 0, 1, 0, 0}, + {&__pyx_n_s_ImportError, __pyx_k_ImportError, sizeof(__pyx_k_ImportError), 0, 0, 1, 1}, + {&__pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_k_MAX_NGRAMS_IN_BATCH, sizeof(__pyx_k_MAX_NGRAMS_IN_BATCH), 0, 0, 1, 1}, + {&__pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_k_MAX_WORDS_IN_BATCH, sizeof(__pyx_k_MAX_WORDS_IN_BATCH), 0, 0, 1, 1}, + {&__pyx_kp_u_Non_native_byte_order_not_suppor, __pyx_k_Non_native_byte_order_not_suppor, sizeof(__pyx_k_Non_native_byte_order_not_suppor), 0, 1, 0, 0}, + {&__pyx_n_s_ONE, __pyx_k_ONE, sizeof(__pyx_k_ONE), 0, 0, 1, 1}, + {&__pyx_n_s_ONEF, __pyx_k_ONEF, sizeof(__pyx_k_ONEF), 0, 0, 1, 1}, + {&__pyx_n_s_RuntimeError, __pyx_k_RuntimeError, sizeof(__pyx_k_RuntimeError), 0, 0, 1, 1}, + {&__pyx_n_s_ValueError, __pyx_k_ValueError, sizeof(__pyx_k_ValueError), 0, 0, 1, 1}, + {&__pyx_n_s_ZERO, __pyx_k_ZERO, sizeof(__pyx_k_ZERO), 0, 0, 1, 1}, + {&__pyx_n_s_ZEROF, __pyx_k_ZEROF, sizeof(__pyx_k_ZEROF), 0, 0, 1, 1}, + {&__pyx_n_s__8, __pyx_k__8, sizeof(__pyx_k__8), 0, 0, 1, 1}, + {&__pyx_n_s_average_inner, __pyx_k_average_inner, sizeof(__pyx_k_average_inner), 0, 0, 1, 1}, + {&__pyx_kp_s_average_inner_pyx, __pyx_k_average_inner_pyx, sizeof(__pyx_k_average_inner_pyx), 0, 0, 1, 0}, + {&__pyx_n_s_bucket, __pyx_k_bucket, sizeof(__pyx_k_bucket), 0, 0, 1, 1}, + {&__pyx_n_s_cline_in_traceback, __pyx_k_cline_in_traceback, sizeof(__pyx_k_cline_in_traceback), 0, 0, 1, 1}, + {&__pyx_n_s_compute_ngrams_bytes, __pyx_k_compute_ngrams_bytes, sizeof(__pyx_k_compute_ngrams_bytes), 0, 0, 1, 1}, + {&__pyx_n_s_cpointer, __pyx_k_cpointer, sizeof(__pyx_k_cpointer), 0, 0, 1, 1}, + {&__pyx_n_s_eff_sentences, __pyx_k_eff_sentences, sizeof(__pyx_k_eff_sentences), 0, 0, 1, 1}, + {&__pyx_n_s_eff_words, __pyx_k_eff_words, sizeof(__pyx_k_eff_words), 0, 0, 1, 1}, + {&__pyx_n_s_enumerate, __pyx_k_enumerate, sizeof(__pyx_k_enumerate), 0, 0, 1, 1}, + {&__pyx_n_s_fblas, __pyx_k_fblas, sizeof(__pyx_k_fblas), 0, 0, 1, 1}, + {&__pyx_n_s_fill, __pyx_k_fill, sizeof(__pyx_k_fill), 0, 0, 1, 1}, + {&__pyx_n_s_ft, __pyx_k_ft, sizeof(__pyx_k_ft), 0, 0, 1, 1}, + {&__pyx_n_s_ft_hash_bytes, __pyx_k_ft_hash_bytes, sizeof(__pyx_k_ft_hash_bytes), 0, 0, 1, 1}, + {&__pyx_n_s_gensim_models__utils_any2vec, __pyx_k_gensim_models__utils_any2vec, sizeof(__pyx_k_gensim_models__utils_any2vec), 0, 0, 1, 1}, + {&__pyx_n_s_import, __pyx_k_import, sizeof(__pyx_k_import), 0, 0, 1, 1}, + {&__pyx_n_s_index, __pyx_k_index, sizeof(__pyx_k_index), 0, 0, 1, 1}, + {&__pyx_n_s_indexed_sentences, __pyx_k_indexed_sentences, sizeof(__pyx_k_indexed_sentences), 0, 0, 1, 1}, + {&__pyx_n_s_init, __pyx_k_init, sizeof(__pyx_k_init), 0, 0, 1, 1}, + {&__pyx_n_s_is_ft, __pyx_k_is_ft, sizeof(__pyx_k_is_ft), 0, 0, 1, 1}, + {&__pyx_n_s_main, __pyx_k_main, sizeof(__pyx_k_main), 0, 0, 1, 1}, + {&__pyx_n_s_max, __pyx_k_max, sizeof(__pyx_k_max), 0, 0, 1, 1}, + {&__pyx_n_s_max_n, __pyx_k_max_n, sizeof(__pyx_k_max_n), 0, 0, 1, 1}, + {&__pyx_n_s_memory, __pyx_k_memory, sizeof(__pyx_k_memory), 0, 0, 1, 1}, + {&__pyx_n_s_min_n, __pyx_k_min_n, sizeof(__pyx_k_min_n), 0, 0, 1, 1}, + {&__pyx_n_s_model, __pyx_k_model, sizeof(__pyx_k_model), 0, 0, 1, 1}, + {&__pyx_n_s_name, __pyx_k_name, sizeof(__pyx_k_name), 0, 0, 1, 1}, + {&__pyx_kp_u_ndarray_is_not_C_contiguous, __pyx_k_ndarray_is_not_C_contiguous, sizeof(__pyx_k_ndarray_is_not_C_contiguous), 0, 1, 0, 0}, + {&__pyx_kp_u_ndarray_is_not_Fortran_contiguou, __pyx_k_ndarray_is_not_Fortran_contiguou, sizeof(__pyx_k_ndarray_is_not_Fortran_contiguou), 0, 1, 0, 0}, + {&__pyx_n_s_np, __pyx_k_np, sizeof(__pyx_k_np), 0, 0, 1, 1}, + {&__pyx_n_s_numpy, __pyx_k_numpy, sizeof(__pyx_k_numpy), 0, 0, 1, 1}, + {&__pyx_kp_s_numpy_core_multiarray_failed_to, __pyx_k_numpy_core_multiarray_failed_to, sizeof(__pyx_k_numpy_core_multiarray_failed_to), 0, 0, 1, 0}, + {&__pyx_kp_s_numpy_core_umath_failed_to_impor, __pyx_k_numpy_core_umath_failed_to_impor, sizeof(__pyx_k_numpy_core_umath_failed_to_impor), 0, 0, 1, 0}, + {&__pyx_n_s_pyx_capi, __pyx_k_pyx_capi, sizeof(__pyx_k_pyx_capi), 0, 0, 1, 1}, + {&__pyx_n_s_range, __pyx_k_range, sizeof(__pyx_k_range), 0, 0, 1, 1}, + {&__pyx_n_s_saxpy, __pyx_k_saxpy, sizeof(__pyx_k_saxpy), 0, 0, 1, 1}, + {&__pyx_n_s_scipy_linalg_blas, __pyx_k_scipy_linalg_blas, sizeof(__pyx_k_scipy_linalg_blas), 0, 0, 1, 1}, + {&__pyx_n_s_sscal, __pyx_k_sscal, sizeof(__pyx_k_sscal), 0, 0, 1, 1}, + {&__pyx_n_s_sv, __pyx_k_sv, sizeof(__pyx_k_sv), 0, 0, 1, 1}, + {&__pyx_n_s_target, __pyx_k_target, sizeof(__pyx_k_target), 0, 0, 1, 1}, + {&__pyx_n_s_test, __pyx_k_test, sizeof(__pyx_k_test), 0, 0, 1, 1}, + {&__pyx_n_s_train_average_cy, __pyx_k_train_average_cy, sizeof(__pyx_k_train_average_cy), 0, 0, 1, 1}, + {&__pyx_kp_u_unknown_dtype_code_in_numpy_pxd, __pyx_k_unknown_dtype_code_in_numpy_pxd, sizeof(__pyx_k_unknown_dtype_code_in_numpy_pxd), 0, 1, 0, 0}, + {&__pyx_n_s_vector_size, __pyx_k_vector_size, sizeof(__pyx_k_vector_size), 0, 0, 1, 1}, + {&__pyx_n_s_vectors, __pyx_k_vectors, sizeof(__pyx_k_vectors), 0, 0, 1, 1}, + {&__pyx_n_s_vectors_ngrams, __pyx_k_vectors_ngrams, sizeof(__pyx_k_vectors_ngrams), 0, 0, 1, 1}, + {&__pyx_n_s_vectors_vocab, __pyx_k_vectors_vocab, sizeof(__pyx_k_vectors_vocab), 0, 0, 1, 1}, + {&__pyx_n_s_vocab, __pyx_k_vocab, sizeof(__pyx_k_vocab), 0, 0, 1, 1}, + {&__pyx_n_s_w2v, __pyx_k_w2v, sizeof(__pyx_k_w2v), 0, 0, 1, 1}, + {&__pyx_n_s_word_weights, __pyx_k_word_weights, sizeof(__pyx_k_word_weights), 0, 0, 1, 1}, + {&__pyx_n_s_workers, __pyx_k_workers, sizeof(__pyx_k_workers), 0, 0, 1, 1}, + {&__pyx_n_s_wv, __pyx_k_wv, sizeof(__pyx_k_wv), 0, 0, 1, 1}, + {0, 0, 0, 0, 0, 0, 0} +}; +static CYTHON_SMALL_CODE int __Pyx_InitCachedBuiltins(void) { + __pyx_builtin_enumerate = __Pyx_GetBuiltinName(__pyx_n_s_enumerate); if (!__pyx_builtin_enumerate) __PYX_ERR(0, 197, __pyx_L1_error) + __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 237, __pyx_L1_error) + __pyx_builtin_ValueError = __Pyx_GetBuiltinName(__pyx_n_s_ValueError); if (!__pyx_builtin_ValueError) __PYX_ERR(1, 272, __pyx_L1_error) + __pyx_builtin_RuntimeError = __Pyx_GetBuiltinName(__pyx_n_s_RuntimeError); if (!__pyx_builtin_RuntimeError) __PYX_ERR(1, 856, __pyx_L1_error) + __pyx_builtin_ImportError = __Pyx_GetBuiltinName(__pyx_n_s_ImportError); if (!__pyx_builtin_ImportError) __PYX_ERR(1, 1038, __pyx_L1_error) + return 0; + __pyx_L1_error:; + return -1; +} + +static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__Pyx_InitCachedConstants", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":272 + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_C_CONTIGUOUS)): + * raise ValueError(u"ndarray is not C contiguous") # <<<<<<<<<<<<<< + * + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) + */ + __pyx_tuple_ = PyTuple_Pack(1, __pyx_kp_u_ndarray_is_not_C_contiguous); if (unlikely(!__pyx_tuple_)) __PYX_ERR(1, 272, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple_); + __Pyx_GIVEREF(__pyx_tuple_); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":276 + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_F_CONTIGUOUS)): + * raise ValueError(u"ndarray is not Fortran contiguous") # <<<<<<<<<<<<<< + * + * info.buf = PyArray_DATA(self) + */ + __pyx_tuple__2 = PyTuple_Pack(1, __pyx_kp_u_ndarray_is_not_Fortran_contiguou); if (unlikely(!__pyx_tuple__2)) __PYX_ERR(1, 276, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__2); + __Pyx_GIVEREF(__pyx_tuple__2); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":306 + * if ((descr.byteorder == c'>' and little_endian) or + * (descr.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") # <<<<<<<<<<<<<< + * if t == NPY_BYTE: f = "b" + * elif t == NPY_UBYTE: f = "B" + */ + __pyx_tuple__3 = PyTuple_Pack(1, __pyx_kp_u_Non_native_byte_order_not_suppor); if (unlikely(!__pyx_tuple__3)) __PYX_ERR(1, 306, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__3); + __Pyx_GIVEREF(__pyx_tuple__3); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":856 + * + * if (end - f) - (new_offset - offset[0]) < 15: + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") # <<<<<<<<<<<<<< + * + * if ((child.byteorder == c'>' and little_endian) or + */ + __pyx_tuple__4 = PyTuple_Pack(1, __pyx_kp_u_Format_string_allocated_too_shor); if (unlikely(!__pyx_tuple__4)) __PYX_ERR(1, 856, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__4); + __Pyx_GIVEREF(__pyx_tuple__4); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":880 + * t = child.type_num + * if end - f < 5: + * raise RuntimeError(u"Format string allocated too short.") # <<<<<<<<<<<<<< + * + * # Until ticket #99 is fixed, use integers to avoid warnings + */ + __pyx_tuple__5 = PyTuple_Pack(1, __pyx_kp_u_Format_string_allocated_too_shor_2); if (unlikely(!__pyx_tuple__5)) __PYX_ERR(1, 880, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__5); + __Pyx_GIVEREF(__pyx_tuple__5); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1038 + * _import_array() + * except Exception: + * raise ImportError("numpy.core.multiarray failed to import") # <<<<<<<<<<<<<< + * + * cdef inline int import_umath() except -1: + */ + __pyx_tuple__6 = PyTuple_Pack(1, __pyx_kp_s_numpy_core_multiarray_failed_to); if (unlikely(!__pyx_tuple__6)) __PYX_ERR(1, 1038, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__6); + __Pyx_GIVEREF(__pyx_tuple__6); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1044 + * _import_umath() + * except Exception: + * raise ImportError("numpy.core.umath failed to import") # <<<<<<<<<<<<<< + * + * cdef inline int import_ufunc() except -1: + */ + __pyx_tuple__7 = PyTuple_Pack(1, __pyx_kp_s_numpy_core_umath_failed_to_impor); if (unlikely(!__pyx_tuple__7)) __PYX_ERR(1, 1044, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__7); + __Pyx_GIVEREF(__pyx_tuple__7); + + /* "average_inner.pyx":313 + * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + * + * def train_average_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< + * """Training on a sequence of sentences and update the target ndarray. + * + */ + __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 313, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__9); + __Pyx_GIVEREF(__pyx_tuple__9); + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_train_average_cy, 313, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 313, __pyx_L1_error) + + /* "average_inner.pyx":359 + * return eff_sentences, eff_words + * + * def init(): # <<<<<<<<<<<<<< + * return 1 + * + */ + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_init, 359, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 359, __pyx_L1_error) + __Pyx_RefNannyFinishContext(); + return 0; + __pyx_L1_error:; + __Pyx_RefNannyFinishContext(); + return -1; +} + +static CYTHON_SMALL_CODE int __Pyx_InitGlobals(void) { + if (__Pyx_InitStrings(__pyx_string_tab) < 0) __PYX_ERR(0, 1, __pyx_L1_error); + __pyx_int_0 = PyInt_FromLong(0); if (unlikely(!__pyx_int_0)) __PYX_ERR(0, 1, __pyx_L1_error) + __pyx_int_1 = PyInt_FromLong(1); if (unlikely(!__pyx_int_1)) __PYX_ERR(0, 1, __pyx_L1_error) + __pyx_int_40 = PyInt_FromLong(40); if (unlikely(!__pyx_int_40)) __PYX_ERR(0, 1, __pyx_L1_error) + __pyx_int_10000 = PyInt_FromLong(10000L); if (unlikely(!__pyx_int_10000)) __PYX_ERR(0, 1, __pyx_L1_error) + return 0; + __pyx_L1_error:; + return -1; +} + +static CYTHON_SMALL_CODE int __Pyx_modinit_global_init_code(void); /*proto*/ +static CYTHON_SMALL_CODE int __Pyx_modinit_variable_export_code(void); /*proto*/ +static CYTHON_SMALL_CODE int __Pyx_modinit_function_export_code(void); /*proto*/ +static CYTHON_SMALL_CODE int __Pyx_modinit_type_init_code(void); /*proto*/ +static CYTHON_SMALL_CODE int __Pyx_modinit_type_import_code(void); /*proto*/ +static CYTHON_SMALL_CODE int __Pyx_modinit_variable_import_code(void); /*proto*/ +static CYTHON_SMALL_CODE int __Pyx_modinit_function_import_code(void); /*proto*/ + +static int __Pyx_modinit_global_init_code(void) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__Pyx_modinit_global_init_code", 0); + /*--- Global init code ---*/ + __Pyx_RefNannyFinishContext(); + return 0; +} + +static int __Pyx_modinit_variable_export_code(void) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__Pyx_modinit_variable_export_code", 0); + /*--- Variable export code ---*/ + if (__Pyx_ExportVoidPtr(__pyx_n_s_saxpy, (void *)&__pyx_v_13average_inner_saxpy, "__pyx_t_13average_inner_saxpy_ptr") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ExportVoidPtr(__pyx_n_s_sscal, (void *)&__pyx_v_13average_inner_sscal, "__pyx_t_13average_inner_sscal_ptr") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ExportVoidPtr(__pyx_n_s_ONE, (void *)&__pyx_v_13average_inner_ONE, "int") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ExportVoidPtr(__pyx_n_s_ZERO, (void *)&__pyx_v_13average_inner_ZERO, "int") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ExportVoidPtr(__pyx_n_s_ONEF, (void *)&__pyx_v_13average_inner_ONEF, "__pyx_t_13average_inner_REAL_t") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ExportVoidPtr(__pyx_n_s_ZEROF, (void *)&__pyx_v_13average_inner_ZEROF, "__pyx_t_13average_inner_REAL_t") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + __Pyx_RefNannyFinishContext(); + return 0; + __pyx_L1_error:; + __Pyx_RefNannyFinishContext(); + return -1; +} + +static int __Pyx_modinit_function_export_code(void) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__Pyx_modinit_function_export_code", 0); + /*--- Function export code ---*/ + if (__Pyx_ExportFunction("init_base_s2v_config", (void (*)(void))__pyx_f_13average_inner_init_base_s2v_config, "PyObject *(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ExportFunction("init_ft_s2v_config", (void (*)(void))__pyx_f_13average_inner_init_ft_s2v_config, "PyObject *(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ExportFunction("populate_base_s2v_config", (void (*)(void))__pyx_f_13average_inner_populate_base_s2v_config, "PyObject *(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ExportFunction("populate_ft_s2v_config", (void (*)(void))__pyx_f_13average_inner_populate_ft_s2v_config, "PyObject *(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + __Pyx_RefNannyFinishContext(); + return 0; + __pyx_L1_error:; + __Pyx_RefNannyFinishContext(); + return -1; +} + +static int __Pyx_modinit_type_init_code(void) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__Pyx_modinit_type_init_code", 0); + /*--- Type init code ---*/ + __Pyx_RefNannyFinishContext(); + return 0; +} + +static int __Pyx_modinit_type_import_code(void) { + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("__Pyx_modinit_type_import_code", 0); + /*--- Type import code ---*/ + __pyx_t_1 = PyImport_ImportModule(__Pyx_BUILTIN_MODULE_NAME); if (unlikely(!__pyx_t_1)) __PYX_ERR(2, 9, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_ptype_7cpython_4type_type = __Pyx_ImportType(__pyx_t_1, __Pyx_BUILTIN_MODULE_NAME, "type", + #if defined(PYPY_VERSION_NUM) && PYPY_VERSION_NUM < 0x050B0000 + sizeof(PyTypeObject), + #else + sizeof(PyHeapTypeObject), + #endif + __Pyx_ImportType_CheckSize_Warn); + if (!__pyx_ptype_7cpython_4type_type) __PYX_ERR(2, 9, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_1 = PyImport_ImportModule("numpy"); if (unlikely(!__pyx_t_1)) __PYX_ERR(1, 206, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_ptype_5numpy_dtype = __Pyx_ImportType(__pyx_t_1, "numpy", "dtype", sizeof(PyArray_Descr), __Pyx_ImportType_CheckSize_Ignore); + if (!__pyx_ptype_5numpy_dtype) __PYX_ERR(1, 206, __pyx_L1_error) + __pyx_ptype_5numpy_flatiter = __Pyx_ImportType(__pyx_t_1, "numpy", "flatiter", sizeof(PyArrayIterObject), __Pyx_ImportType_CheckSize_Warn); + if (!__pyx_ptype_5numpy_flatiter) __PYX_ERR(1, 229, __pyx_L1_error) + __pyx_ptype_5numpy_broadcast = __Pyx_ImportType(__pyx_t_1, "numpy", "broadcast", sizeof(PyArrayMultiIterObject), __Pyx_ImportType_CheckSize_Warn); + if (!__pyx_ptype_5numpy_broadcast) __PYX_ERR(1, 233, __pyx_L1_error) + __pyx_ptype_5numpy_ndarray = __Pyx_ImportType(__pyx_t_1, "numpy", "ndarray", sizeof(PyArrayObject), __Pyx_ImportType_CheckSize_Ignore); + if (!__pyx_ptype_5numpy_ndarray) __PYX_ERR(1, 242, __pyx_L1_error) + __pyx_ptype_5numpy_ufunc = __Pyx_ImportType(__pyx_t_1, "numpy", "ufunc", sizeof(PyUFuncObject), __Pyx_ImportType_CheckSize_Warn); + if (!__pyx_ptype_5numpy_ufunc) __PYX_ERR(1, 918, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __Pyx_RefNannyFinishContext(); + return 0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_RefNannyFinishContext(); + return -1; +} + +static int __Pyx_modinit_variable_import_code(void) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__Pyx_modinit_variable_import_code", 0); + /*--- Variable import code ---*/ + __Pyx_RefNannyFinishContext(); + return 0; +} + +static int __Pyx_modinit_function_import_code(void) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__Pyx_modinit_function_import_code", 0); + /*--- Function import code ---*/ + __Pyx_RefNannyFinishContext(); + return 0; +} + + +#if PY_MAJOR_VERSION < 3 +#ifdef CYTHON_NO_PYINIT_EXPORT +#define __Pyx_PyMODINIT_FUNC void +#else +#define __Pyx_PyMODINIT_FUNC PyMODINIT_FUNC +#endif +#else +#ifdef CYTHON_NO_PYINIT_EXPORT +#define __Pyx_PyMODINIT_FUNC PyObject * +#else +#define __Pyx_PyMODINIT_FUNC PyMODINIT_FUNC +#endif +#endif + + +#if PY_MAJOR_VERSION < 3 +__Pyx_PyMODINIT_FUNC initaverage_inner(void) CYTHON_SMALL_CODE; /*proto*/ +__Pyx_PyMODINIT_FUNC initaverage_inner(void) +#else +__Pyx_PyMODINIT_FUNC PyInit_average_inner(void) CYTHON_SMALL_CODE; /*proto*/ +__Pyx_PyMODINIT_FUNC PyInit_average_inner(void) +#if CYTHON_PEP489_MULTI_PHASE_INIT +{ + return PyModuleDef_Init(&__pyx_moduledef); +} +static CYTHON_SMALL_CODE int __Pyx_check_single_interpreter(void) { + #if PY_VERSION_HEX >= 0x030700A1 + static PY_INT64_T main_interpreter_id = -1; + PY_INT64_T current_id = PyInterpreterState_GetID(PyThreadState_Get()->interp); + if (main_interpreter_id == -1) { + main_interpreter_id = current_id; + return (unlikely(current_id == -1)) ? -1 : 0; + } else if (unlikely(main_interpreter_id != current_id)) + #else + static PyInterpreterState *main_interpreter = NULL; + PyInterpreterState *current_interpreter = PyThreadState_Get()->interp; + if (!main_interpreter) { + main_interpreter = current_interpreter; + } else if (unlikely(main_interpreter != current_interpreter)) + #endif + { + PyErr_SetString( + PyExc_ImportError, + "Interpreter change detected - this module can only be loaded into one interpreter per process."); + return -1; + } + return 0; +} +static CYTHON_SMALL_CODE int __Pyx_copy_spec_to_module(PyObject *spec, PyObject *moddict, const char* from_name, const char* to_name, int allow_none) { + PyObject *value = PyObject_GetAttrString(spec, from_name); + int result = 0; + if (likely(value)) { + if (allow_none || value != Py_None) { + result = PyDict_SetItemString(moddict, to_name, value); + } + Py_DECREF(value); + } else if (PyErr_ExceptionMatches(PyExc_AttributeError)) { + PyErr_Clear(); + } else { + result = -1; + } + return result; +} +static CYTHON_SMALL_CODE PyObject* __pyx_pymod_create(PyObject *spec, CYTHON_UNUSED PyModuleDef *def) { + PyObject *module = NULL, *moddict, *modname; + if (__Pyx_check_single_interpreter()) + return NULL; + if (__pyx_m) + return __Pyx_NewRef(__pyx_m); + modname = PyObject_GetAttrString(spec, "name"); + if (unlikely(!modname)) goto bad; + module = PyModule_NewObject(modname); + Py_DECREF(modname); + if (unlikely(!module)) goto bad; + moddict = PyModule_GetDict(module); + if (unlikely(!moddict)) goto bad; + if (unlikely(__Pyx_copy_spec_to_module(spec, moddict, "loader", "__loader__", 1) < 0)) goto bad; + if (unlikely(__Pyx_copy_spec_to_module(spec, moddict, "origin", "__file__", 1) < 0)) goto bad; + if (unlikely(__Pyx_copy_spec_to_module(spec, moddict, "parent", "__package__", 1) < 0)) goto bad; + if (unlikely(__Pyx_copy_spec_to_module(spec, moddict, "submodule_search_locations", "__path__", 0) < 0)) goto bad; + return module; +bad: + Py_XDECREF(module); + return NULL; +} + + +static CYTHON_SMALL_CODE int __pyx_pymod_exec_average_inner(PyObject *__pyx_pyinit_module) +#endif +#endif +{ + PyObject *__pyx_t_1 = NULL; + PyObject *__pyx_t_2 = NULL; + __Pyx_RefNannyDeclarations + #if CYTHON_PEP489_MULTI_PHASE_INIT + if (__pyx_m) { + if (__pyx_m == __pyx_pyinit_module) return 0; + PyErr_SetString(PyExc_RuntimeError, "Module 'average_inner' has already been imported. Re-initialisation is not supported."); + return -1; + } + #elif PY_MAJOR_VERSION >= 3 + if (__pyx_m) return __Pyx_NewRef(__pyx_m); + #endif + #if CYTHON_REFNANNY +__Pyx_RefNanny = __Pyx_RefNannyImportAPI("refnanny"); +if (!__Pyx_RefNanny) { + PyErr_Clear(); + __Pyx_RefNanny = __Pyx_RefNannyImportAPI("Cython.Runtime.refnanny"); + if (!__Pyx_RefNanny) + Py_FatalError("failed to import 'refnanny' module"); +} +#endif + __Pyx_RefNannySetupContext("__Pyx_PyMODINIT_FUNC PyInit_average_inner(void)", 0); + if (__Pyx_check_binary_version() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #ifdef __Pxy_PyFrame_Initialize_Offsets + __Pxy_PyFrame_Initialize_Offsets(); + #endif + __pyx_empty_tuple = PyTuple_New(0); if (unlikely(!__pyx_empty_tuple)) __PYX_ERR(0, 1, __pyx_L1_error) + __pyx_empty_bytes = PyBytes_FromStringAndSize("", 0); if (unlikely(!__pyx_empty_bytes)) __PYX_ERR(0, 1, __pyx_L1_error) + __pyx_empty_unicode = PyUnicode_FromStringAndSize("", 0); if (unlikely(!__pyx_empty_unicode)) __PYX_ERR(0, 1, __pyx_L1_error) + #ifdef __Pyx_CyFunction_USED + if (__pyx_CyFunction_init() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + #ifdef __Pyx_FusedFunction_USED + if (__pyx_FusedFunction_init() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + #ifdef __Pyx_Coroutine_USED + if (__pyx_Coroutine_init() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + #ifdef __Pyx_Generator_USED + if (__pyx_Generator_init() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + #ifdef __Pyx_AsyncGen_USED + if (__pyx_AsyncGen_init() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + #ifdef __Pyx_StopAsyncIteration_USED + if (__pyx_StopAsyncIteration_init() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + /*--- Library function declarations ---*/ + /*--- Threads initialization code ---*/ + #if defined(__PYX_FORCE_INIT_THREADS) && __PYX_FORCE_INIT_THREADS + #ifdef WITH_THREAD /* Python build with threading support? */ + PyEval_InitThreads(); + #endif + #endif + /*--- Module creation code ---*/ + #if CYTHON_PEP489_MULTI_PHASE_INIT + __pyx_m = __pyx_pyinit_module; + Py_INCREF(__pyx_m); + #else + #if PY_MAJOR_VERSION < 3 + __pyx_m = Py_InitModule4("average_inner", __pyx_methods, __pyx_k_Optimized_cython_functions_for_c, 0, PYTHON_API_VERSION); Py_XINCREF(__pyx_m); + #else + __pyx_m = PyModule_Create(&__pyx_moduledef); + #endif + if (unlikely(!__pyx_m)) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + __pyx_d = PyModule_GetDict(__pyx_m); if (unlikely(!__pyx_d)) __PYX_ERR(0, 1, __pyx_L1_error) + Py_INCREF(__pyx_d); + __pyx_b = PyImport_AddModule(__Pyx_BUILTIN_MODULE_NAME); if (unlikely(!__pyx_b)) __PYX_ERR(0, 1, __pyx_L1_error) + Py_INCREF(__pyx_b); + __pyx_cython_runtime = PyImport_AddModule((char *) "cython_runtime"); if (unlikely(!__pyx_cython_runtime)) __PYX_ERR(0, 1, __pyx_L1_error) + Py_INCREF(__pyx_cython_runtime); + if (PyObject_SetAttrString(__pyx_m, "__builtins__", __pyx_b) < 0) __PYX_ERR(0, 1, __pyx_L1_error); + /*--- Initialize various global constants etc. ---*/ + if (__Pyx_InitGlobals() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #if PY_MAJOR_VERSION < 3 && (__PYX_DEFAULT_STRING_ENCODING_IS_ASCII || __PYX_DEFAULT_STRING_ENCODING_IS_DEFAULT) + if (__Pyx_init_sys_getdefaultencoding_params() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + if (__pyx_module_is_main_average_inner) { + if (PyObject_SetAttr(__pyx_m, __pyx_n_s_name, __pyx_n_s_main) < 0) __PYX_ERR(0, 1, __pyx_L1_error) + } + #if PY_MAJOR_VERSION >= 3 + { + PyObject *modules = PyImport_GetModuleDict(); if (unlikely(!modules)) __PYX_ERR(0, 1, __pyx_L1_error) + if (!PyDict_GetItemString(modules, "average_inner")) { + if (unlikely(PyDict_SetItemString(modules, "average_inner", __pyx_m) < 0)) __PYX_ERR(0, 1, __pyx_L1_error) + } + } + #endif + /*--- Builtin init code ---*/ + if (__Pyx_InitCachedBuiltins() < 0) goto __pyx_L1_error; + /*--- Constants init code ---*/ + if (__Pyx_InitCachedConstants() < 0) goto __pyx_L1_error; + /*--- Global type/function init code ---*/ + (void)__Pyx_modinit_global_init_code(); + if (unlikely(__Pyx_modinit_variable_export_code() != 0)) goto __pyx_L1_error; + if (unlikely(__Pyx_modinit_function_export_code() != 0)) goto __pyx_L1_error; + (void)__Pyx_modinit_type_init_code(); + if (unlikely(__Pyx_modinit_type_import_code() != 0)) goto __pyx_L1_error; + (void)__Pyx_modinit_variable_import_code(); + (void)__Pyx_modinit_function_import_code(); + /*--- Execution code ---*/ + #if defined(__Pyx_Generator_USED) || defined(__Pyx_Coroutine_USED) + if (__Pyx_patch_abc() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + + /* "average_inner.pyx":14 + * + * import cython + * import numpy as np # <<<<<<<<<<<<<< + * + * cimport numpy as np + */ + __pyx_t_1 = __Pyx_Import(__pyx_n_s_numpy, 0, -1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 14, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + if (PyDict_SetItem(__pyx_d, __pyx_n_s_np, __pyx_t_1) < 0) __PYX_ERR(0, 14, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":18 + * cimport numpy as np + * + * from gensim.models._utils_any2vec import compute_ngrams_bytes, ft_hash_bytes # <<<<<<<<<<<<<< + * + * from libc.string cimport memset + */ + __pyx_t_1 = PyList_New(2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 18, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_INCREF(__pyx_n_s_compute_ngrams_bytes); + __Pyx_GIVEREF(__pyx_n_s_compute_ngrams_bytes); + PyList_SET_ITEM(__pyx_t_1, 0, __pyx_n_s_compute_ngrams_bytes); + __Pyx_INCREF(__pyx_n_s_ft_hash_bytes); + __Pyx_GIVEREF(__pyx_n_s_ft_hash_bytes); + PyList_SET_ITEM(__pyx_t_1, 1, __pyx_n_s_ft_hash_bytes); + __pyx_t_2 = __Pyx_Import(__pyx_n_s_gensim_models__utils_any2vec, __pyx_t_1, -1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 18, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_2); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_1 = __Pyx_ImportFrom(__pyx_t_2, __pyx_n_s_compute_ngrams_bytes); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 18, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + if (PyDict_SetItem(__pyx_d, __pyx_n_s_compute_ngrams_bytes, __pyx_t_1) < 0) __PYX_ERR(0, 18, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_1 = __Pyx_ImportFrom(__pyx_t_2, __pyx_n_s_ft_hash_bytes); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 18, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + if (PyDict_SetItem(__pyx_d, __pyx_n_s_ft_hash_bytes, __pyx_t_1) < 0) __PYX_ERR(0, 18, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; + + /* "average_inner.pyx":22 + * from libc.string cimport memset + * + * import scipy.linalg.blas as fblas # <<<<<<<<<<<<<< + * + * cdef saxpy_ptr saxpy=PyCObject_AsVoidPtr(fblas.saxpy._cpointer) # y += alpha * x + */ + __pyx_t_2 = PyList_New(1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 22, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_2); + __Pyx_INCREF(__pyx_n_s__8); + __Pyx_GIVEREF(__pyx_n_s__8); + PyList_SET_ITEM(__pyx_t_2, 0, __pyx_n_s__8); + __pyx_t_1 = __Pyx_Import(__pyx_n_s_scipy_linalg_blas, __pyx_t_2, -1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 22, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; + if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_1) < 0) __PYX_ERR(0, 22, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":24 + * import scipy.linalg.blas as fblas + * + * cdef saxpy_ptr saxpy=PyCObject_AsVoidPtr(fblas.saxpy._cpointer) # y += alpha * x # <<<<<<<<<<<<<< + * cdef sscal_ptr sscal=PyCObject_AsVoidPtr(fblas.sscal._cpointer) # x = alpha * x + * + */ + __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_fblas); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 24, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_saxpy); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 24, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_2); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_2, __pyx_n_s_cpointer); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 24, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; + __pyx_v_13average_inner_saxpy = ((__pyx_t_13average_inner_saxpy_ptr)PyCObject_AsVoidPtr(__pyx_t_1)); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":25 + * + * cdef saxpy_ptr saxpy=PyCObject_AsVoidPtr(fblas.saxpy._cpointer) # y += alpha * x + * cdef sscal_ptr sscal=PyCObject_AsVoidPtr(fblas.sscal._cpointer) # x = alpha * x # <<<<<<<<<<<<<< + * + * cdef int ONE = 1 + */ + __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_fblas); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 25, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_sscal); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 25, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_2); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_2, __pyx_n_s_cpointer); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 25, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; + __pyx_v_13average_inner_sscal = ((__pyx_t_13average_inner_sscal_ptr)PyCObject_AsVoidPtr(__pyx_t_1)); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":27 + * cdef sscal_ptr sscal=PyCObject_AsVoidPtr(fblas.sscal._cpointer) # x = alpha * x + * + * cdef int ONE = 1 # <<<<<<<<<<<<<< + * cdef int ZERO = 0 + * + */ + __pyx_v_13average_inner_ONE = ((int)1); + + /* "average_inner.pyx":28 + * + * cdef int ONE = 1 + * cdef int ZERO = 0 # <<<<<<<<<<<<<< + * + * cdef REAL_t ONEF = 1.0 + */ + __pyx_v_13average_inner_ZERO = ((int)0); + + /* "average_inner.pyx":30 + * cdef int ZERO = 0 + * + * cdef REAL_t ONEF = 1.0 # <<<<<<<<<<<<<< + * cdef REAL_t ZEROF = 0.0 + * + */ + __pyx_v_13average_inner_ONEF = ((__pyx_t_13average_inner_REAL_t)1.0); + + /* "average_inner.pyx":31 + * + * cdef REAL_t ONEF = 1.0 + * cdef REAL_t ZEROF = 0.0 # <<<<<<<<<<<<<< + * + * DEF MAX_WORDS = 10000 + */ + __pyx_v_13average_inner_ZEROF = ((__pyx_t_13average_inner_REAL_t)0.0); + + /* "average_inner.pyx":313 + * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + * + * def train_average_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< + * """Training on a sequence of sentences and update the target ndarray. + * + */ + __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_1train_average_cy, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 313, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_average_cy, __pyx_t_1) < 0) __PYX_ERR(0, 313, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":359 + * return eff_sentences, eff_words + * + * def init(): # <<<<<<<<<<<<<< + * return 1 + * + */ + __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_3init, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 359, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_1) < 0) __PYX_ERR(0, 359, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "average_inner.pyx":362 + * return 1 + * + * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< + * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS + * FAST_VERSION = init() + */ + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 362, __pyx_L1_error) + + /* "average_inner.pyx":363 + * + * MAX_WORDS_IN_BATCH = MAX_WORDS + * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< + * FAST_VERSION = init() + */ + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 363, __pyx_L1_error) + + /* "average_inner.pyx":364 + * MAX_WORDS_IN_BATCH = MAX_WORDS + * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS + * FAST_VERSION = init() # <<<<<<<<<<<<<< + */ + __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_init); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 364, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_2 = __Pyx_PyObject_CallNoArg(__pyx_t_1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 364, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_2); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_2) < 0) __PYX_ERR(0, 364, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; + + /* "average_inner.pyx":1 + * #!/usr/bin/env cython # <<<<<<<<<<<<<< + * # cython: boundscheck=False + * # cython: wraparound=False + */ + __pyx_t_2 = __Pyx_PyDict_NewPresized(0); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 1, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_2); + if (PyDict_SetItem(__pyx_d, __pyx_n_s_test, __pyx_t_2) < 0) __PYX_ERR(0, 1, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1046 + * raise ImportError("numpy.core.umath failed to import") + * + * cdef inline int import_ufunc() except -1: # <<<<<<<<<<<<<< + * try: + * _import_umath() + */ + + /*--- Wrapped vars code ---*/ + + goto __pyx_L0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_XDECREF(__pyx_t_2); + if (__pyx_m) { + if (__pyx_d) { + __Pyx_AddTraceback("init average_inner", __pyx_clineno, __pyx_lineno, __pyx_filename); + } + Py_CLEAR(__pyx_m); + } else if (!PyErr_Occurred()) { + PyErr_SetString(PyExc_ImportError, "init average_inner"); + } + __pyx_L0:; + __Pyx_RefNannyFinishContext(); + #if CYTHON_PEP489_MULTI_PHASE_INIT + return (__pyx_m != NULL) ? 0 : -1; + #elif PY_MAJOR_VERSION >= 3 + return __pyx_m; + #else + return; + #endif +} + +/* --- Runtime support code --- */ +/* Refnanny */ +#if CYTHON_REFNANNY +static __Pyx_RefNannyAPIStruct *__Pyx_RefNannyImportAPI(const char *modname) { + PyObject *m = NULL, *p = NULL; + void *r = NULL; + m = PyImport_ImportModule(modname); + if (!m) goto end; + p = PyObject_GetAttrString(m, "RefNannyAPI"); + if (!p) goto end; + r = PyLong_AsVoidPtr(p); +end: + Py_XDECREF(p); + Py_XDECREF(m); + return (__Pyx_RefNannyAPIStruct *)r; +} +#endif + +/* PyObjectGetAttrStr */ +#if CYTHON_USE_TYPE_SLOTS +static CYTHON_INLINE PyObject* __Pyx_PyObject_GetAttrStr(PyObject* obj, PyObject* attr_name) { + PyTypeObject* tp = Py_TYPE(obj); + if (likely(tp->tp_getattro)) + return tp->tp_getattro(obj, attr_name); +#if PY_MAJOR_VERSION < 3 + if (likely(tp->tp_getattr)) + return tp->tp_getattr(obj, PyString_AS_STRING(attr_name)); +#endif + return PyObject_GetAttr(obj, attr_name); +} +#endif + +/* GetBuiltinName */ +static PyObject *__Pyx_GetBuiltinName(PyObject *name) { + PyObject* result = __Pyx_PyObject_GetAttrStr(__pyx_b, name); + if (unlikely(!result)) { + PyErr_Format(PyExc_NameError, +#if PY_MAJOR_VERSION >= 3 + "name '%U' is not defined", name); +#else + "name '%.200s' is not defined", PyString_AS_STRING(name)); +#endif + } + return result; +} + +/* GetItemInt */ +static PyObject *__Pyx_GetItemInt_Generic(PyObject *o, PyObject* j) { + PyObject *r; + if (!j) return NULL; + r = PyObject_GetItem(o, j); + Py_DECREF(j); + return r; +} +static CYTHON_INLINE PyObject *__Pyx_GetItemInt_List_Fast(PyObject *o, Py_ssize_t i, + CYTHON_NCP_UNUSED int wraparound, + CYTHON_NCP_UNUSED int boundscheck) { +#if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + Py_ssize_t wrapped_i = i; + if (wraparound & unlikely(i < 0)) { + wrapped_i += PyList_GET_SIZE(o); + } + if ((!boundscheck) || likely(__Pyx_is_valid_index(wrapped_i, PyList_GET_SIZE(o)))) { + PyObject *r = PyList_GET_ITEM(o, wrapped_i); + Py_INCREF(r); + return r; + } + return __Pyx_GetItemInt_Generic(o, PyInt_FromSsize_t(i)); +#else + return PySequence_GetItem(o, i); +#endif +} +static CYTHON_INLINE PyObject *__Pyx_GetItemInt_Tuple_Fast(PyObject *o, Py_ssize_t i, + CYTHON_NCP_UNUSED int wraparound, + CYTHON_NCP_UNUSED int boundscheck) { +#if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + Py_ssize_t wrapped_i = i; + if (wraparound & unlikely(i < 0)) { + wrapped_i += PyTuple_GET_SIZE(o); + } + if ((!boundscheck) || likely(__Pyx_is_valid_index(wrapped_i, PyTuple_GET_SIZE(o)))) { + PyObject *r = PyTuple_GET_ITEM(o, wrapped_i); + Py_INCREF(r); + return r; + } + return __Pyx_GetItemInt_Generic(o, PyInt_FromSsize_t(i)); +#else + return PySequence_GetItem(o, i); +#endif +} +static CYTHON_INLINE PyObject *__Pyx_GetItemInt_Fast(PyObject *o, Py_ssize_t i, int is_list, + CYTHON_NCP_UNUSED int wraparound, + CYTHON_NCP_UNUSED int boundscheck) { +#if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS && CYTHON_USE_TYPE_SLOTS + if (is_list || PyList_CheckExact(o)) { + Py_ssize_t n = ((!wraparound) | likely(i >= 0)) ? i : i + PyList_GET_SIZE(o); + if ((!boundscheck) || (likely(__Pyx_is_valid_index(n, PyList_GET_SIZE(o))))) { + PyObject *r = PyList_GET_ITEM(o, n); + Py_INCREF(r); + return r; + } + } + else if (PyTuple_CheckExact(o)) { + Py_ssize_t n = ((!wraparound) | likely(i >= 0)) ? i : i + PyTuple_GET_SIZE(o); + if ((!boundscheck) || likely(__Pyx_is_valid_index(n, PyTuple_GET_SIZE(o)))) { + PyObject *r = PyTuple_GET_ITEM(o, n); + Py_INCREF(r); + return r; + } + } else { + PySequenceMethods *m = Py_TYPE(o)->tp_as_sequence; + if (likely(m && m->sq_item)) { + if (wraparound && unlikely(i < 0) && likely(m->sq_length)) { + Py_ssize_t l = m->sq_length(o); + if (likely(l >= 0)) { + i += l; + } else { + if (!PyErr_ExceptionMatches(PyExc_OverflowError)) + return NULL; + PyErr_Clear(); + } + } + return m->sq_item(o, i); + } + } +#else + if (is_list || PySequence_Check(o)) { + return PySequence_GetItem(o, i); + } +#endif + return __Pyx_GetItemInt_Generic(o, PyInt_FromSsize_t(i)); +} + +/* ExtTypeTest */ +static CYTHON_INLINE int __Pyx_TypeTest(PyObject *obj, PyTypeObject *type) { + if (unlikely(!type)) { + PyErr_SetString(PyExc_SystemError, "Missing type object"); + return 0; + } + if (likely(__Pyx_TypeCheck(obj, type))) + return 1; + PyErr_Format(PyExc_TypeError, "Cannot convert %.200s to %.200s", + Py_TYPE(obj)->tp_name, type->tp_name); + return 0; +} + +/* PyDictVersioning */ +#if CYTHON_USE_DICT_VERSIONS && CYTHON_USE_TYPE_SLOTS +static CYTHON_INLINE PY_UINT64_T __Pyx_get_tp_dict_version(PyObject *obj) { + PyObject *dict = Py_TYPE(obj)->tp_dict; + return likely(dict) ? __PYX_GET_DICT_VERSION(dict) : 0; +} +static CYTHON_INLINE PY_UINT64_T __Pyx_get_object_dict_version(PyObject *obj) { + PyObject **dictptr = NULL; + Py_ssize_t offset = Py_TYPE(obj)->tp_dictoffset; + if (offset) { +#if CYTHON_COMPILING_IN_CPYTHON + dictptr = (likely(offset > 0)) ? (PyObject **) ((char *)obj + offset) : _PyObject_GetDictPtr(obj); +#else + dictptr = _PyObject_GetDictPtr(obj); +#endif + } + return (dictptr && *dictptr) ? __PYX_GET_DICT_VERSION(*dictptr) : 0; +} +static CYTHON_INLINE int __Pyx_object_dict_version_matches(PyObject* obj, PY_UINT64_T tp_dict_version, PY_UINT64_T obj_dict_version) { + PyObject *dict = Py_TYPE(obj)->tp_dict; + if (unlikely(!dict) || unlikely(tp_dict_version != __PYX_GET_DICT_VERSION(dict))) + return 0; + return obj_dict_version == __Pyx_get_object_dict_version(obj); +} +#endif + +/* GetModuleGlobalName */ +#if CYTHON_USE_DICT_VERSIONS +static PyObject *__Pyx__GetModuleGlobalName(PyObject *name, PY_UINT64_T *dict_version, PyObject **dict_cached_value) +#else +static CYTHON_INLINE PyObject *__Pyx__GetModuleGlobalName(PyObject *name) +#endif +{ + PyObject *result; +#if !CYTHON_AVOID_BORROWED_REFS +#if CYTHON_COMPILING_IN_CPYTHON && PY_VERSION_HEX >= 0x030500A1 + result = _PyDict_GetItem_KnownHash(__pyx_d, name, ((PyASCIIObject *) name)->hash); + __PYX_UPDATE_DICT_CACHE(__pyx_d, result, *dict_cached_value, *dict_version) + if (likely(result)) { + return __Pyx_NewRef(result); + } else if (unlikely(PyErr_Occurred())) { + return NULL; + } +#else + result = PyDict_GetItem(__pyx_d, name); + __PYX_UPDATE_DICT_CACHE(__pyx_d, result, *dict_cached_value, *dict_version) + if (likely(result)) { + return __Pyx_NewRef(result); + } +#endif +#else + result = PyObject_GetItem(__pyx_d, name); + __PYX_UPDATE_DICT_CACHE(__pyx_d, result, *dict_cached_value, *dict_version) + if (likely(result)) { + return __Pyx_NewRef(result); + } + PyErr_Clear(); +#endif + return __Pyx_GetBuiltinName(name); +} + +/* PyCFunctionFastCall */ +#if CYTHON_FAST_PYCCALL +static CYTHON_INLINE PyObject * __Pyx_PyCFunction_FastCall(PyObject *func_obj, PyObject **args, Py_ssize_t nargs) { + PyCFunctionObject *func = (PyCFunctionObject*)func_obj; + PyCFunction meth = PyCFunction_GET_FUNCTION(func); + PyObject *self = PyCFunction_GET_SELF(func); + int flags = PyCFunction_GET_FLAGS(func); + assert(PyCFunction_Check(func)); + assert(METH_FASTCALL == (flags & ~(METH_CLASS | METH_STATIC | METH_COEXIST | METH_KEYWORDS | METH_STACKLESS))); + assert(nargs >= 0); + assert(nargs == 0 || args != NULL); + /* _PyCFunction_FastCallDict() must not be called with an exception set, + because it may clear it (directly or indirectly) and so the + caller loses its exception */ + assert(!PyErr_Occurred()); + if ((PY_VERSION_HEX < 0x030700A0) || unlikely(flags & METH_KEYWORDS)) { + return (*((__Pyx_PyCFunctionFastWithKeywords)(void*)meth)) (self, args, nargs, NULL); + } else { + return (*((__Pyx_PyCFunctionFast)(void*)meth)) (self, args, nargs); + } +} +#endif + +/* PyFunctionFastCall */ +#if CYTHON_FAST_PYCALL +static PyObject* __Pyx_PyFunction_FastCallNoKw(PyCodeObject *co, PyObject **args, Py_ssize_t na, + PyObject *globals) { + PyFrameObject *f; + PyThreadState *tstate = __Pyx_PyThreadState_Current; + PyObject **fastlocals; + Py_ssize_t i; + PyObject *result; + assert(globals != NULL); + /* XXX Perhaps we should create a specialized + PyFrame_New() that doesn't take locals, but does + take builtins without sanity checking them. + */ + assert(tstate != NULL); + f = PyFrame_New(tstate, co, globals, NULL); + if (f == NULL) { + return NULL; + } + fastlocals = __Pyx_PyFrame_GetLocalsplus(f); + for (i = 0; i < na; i++) { + Py_INCREF(*args); + fastlocals[i] = *args++; + } + result = PyEval_EvalFrameEx(f,0); + ++tstate->recursion_depth; + Py_DECREF(f); + --tstate->recursion_depth; + return result; +} +#if 1 || PY_VERSION_HEX < 0x030600B1 +static PyObject *__Pyx_PyFunction_FastCallDict(PyObject *func, PyObject **args, Py_ssize_t nargs, PyObject *kwargs) { + PyCodeObject *co = (PyCodeObject *)PyFunction_GET_CODE(func); + PyObject *globals = PyFunction_GET_GLOBALS(func); + PyObject *argdefs = PyFunction_GET_DEFAULTS(func); + PyObject *closure; +#if PY_MAJOR_VERSION >= 3 + PyObject *kwdefs; +#endif + PyObject *kwtuple, **k; + PyObject **d; + Py_ssize_t nd; + Py_ssize_t nk; + PyObject *result; + assert(kwargs == NULL || PyDict_Check(kwargs)); + nk = kwargs ? PyDict_Size(kwargs) : 0; + if (Py_EnterRecursiveCall((char*)" while calling a Python object")) { + return NULL; + } + if ( +#if PY_MAJOR_VERSION >= 3 + co->co_kwonlyargcount == 0 && +#endif + likely(kwargs == NULL || nk == 0) && + co->co_flags == (CO_OPTIMIZED | CO_NEWLOCALS | CO_NOFREE)) { + if (argdefs == NULL && co->co_argcount == nargs) { + result = __Pyx_PyFunction_FastCallNoKw(co, args, nargs, globals); + goto done; + } + else if (nargs == 0 && argdefs != NULL + && co->co_argcount == Py_SIZE(argdefs)) { + /* function called with no arguments, but all parameters have + a default value: use default values as arguments .*/ + args = &PyTuple_GET_ITEM(argdefs, 0); + result =__Pyx_PyFunction_FastCallNoKw(co, args, Py_SIZE(argdefs), globals); + goto done; + } + } + if (kwargs != NULL) { + Py_ssize_t pos, i; + kwtuple = PyTuple_New(2 * nk); + if (kwtuple == NULL) { + result = NULL; + goto done; + } + k = &PyTuple_GET_ITEM(kwtuple, 0); + pos = i = 0; + while (PyDict_Next(kwargs, &pos, &k[i], &k[i+1])) { + Py_INCREF(k[i]); + Py_INCREF(k[i+1]); + i += 2; + } + nk = i / 2; + } + else { + kwtuple = NULL; + k = NULL; + } + closure = PyFunction_GET_CLOSURE(func); +#if PY_MAJOR_VERSION >= 3 + kwdefs = PyFunction_GET_KW_DEFAULTS(func); +#endif + if (argdefs != NULL) { + d = &PyTuple_GET_ITEM(argdefs, 0); + nd = Py_SIZE(argdefs); + } + else { + d = NULL; + nd = 0; + } +#if PY_MAJOR_VERSION >= 3 + result = PyEval_EvalCodeEx((PyObject*)co, globals, (PyObject *)NULL, + args, (int)nargs, + k, (int)nk, + d, (int)nd, kwdefs, closure); +#else + result = PyEval_EvalCodeEx(co, globals, (PyObject *)NULL, + args, (int)nargs, + k, (int)nk, + d, (int)nd, closure); +#endif + Py_XDECREF(kwtuple); +done: + Py_LeaveRecursiveCall(); + return result; +} +#endif +#endif + +/* PyObjectCall */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE PyObject* __Pyx_PyObject_Call(PyObject *func, PyObject *arg, PyObject *kw) { + PyObject *result; + ternaryfunc call = func->ob_type->tp_call; + if (unlikely(!call)) + return PyObject_Call(func, arg, kw); + if (unlikely(Py_EnterRecursiveCall((char*)" while calling a Python object"))) + return NULL; + result = (*call)(func, arg, kw); + Py_LeaveRecursiveCall(); + if (unlikely(!result) && unlikely(!PyErr_Occurred())) { + PyErr_SetString( + PyExc_SystemError, + "NULL result without error in PyObject_Call"); + } + return result; +} +#endif + +/* PyObjectCall2Args */ +static CYTHON_UNUSED PyObject* __Pyx_PyObject_Call2Args(PyObject* function, PyObject* arg1, PyObject* arg2) { + PyObject *args, *result = NULL; + #if CYTHON_FAST_PYCALL + if (PyFunction_Check(function)) { + PyObject *args[2] = {arg1, arg2}; + return __Pyx_PyFunction_FastCall(function, args, 2); + } + #endif + #if CYTHON_FAST_PYCCALL + if (__Pyx_PyFastCFunction_Check(function)) { + PyObject *args[2] = {arg1, arg2}; + return __Pyx_PyCFunction_FastCall(function, args, 2); + } + #endif + args = PyTuple_New(2); + if (unlikely(!args)) goto done; + Py_INCREF(arg1); + PyTuple_SET_ITEM(args, 0, arg1); + Py_INCREF(arg2); + PyTuple_SET_ITEM(args, 1, arg2); + Py_INCREF(function); + result = __Pyx_PyObject_Call(function, args, NULL); + Py_DECREF(args); + Py_DECREF(function); +done: + return result; +} + +/* PyObjectCallMethO */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallMethO(PyObject *func, PyObject *arg) { + PyObject *self, *result; + PyCFunction cfunc; + cfunc = PyCFunction_GET_FUNCTION(func); + self = PyCFunction_GET_SELF(func); + if (unlikely(Py_EnterRecursiveCall((char*)" while calling a Python object"))) + return NULL; + result = cfunc(self, arg); + Py_LeaveRecursiveCall(); + if (unlikely(!result) && unlikely(!PyErr_Occurred())) { + PyErr_SetString( + PyExc_SystemError, + "NULL result without error in PyObject_Call"); + } + return result; +} +#endif + +/* PyObjectCallOneArg */ +#if CYTHON_COMPILING_IN_CPYTHON +static PyObject* __Pyx__PyObject_CallOneArg(PyObject *func, PyObject *arg) { + PyObject *result; + PyObject *args = PyTuple_New(1); + if (unlikely(!args)) return NULL; + Py_INCREF(arg); + PyTuple_SET_ITEM(args, 0, arg); + result = __Pyx_PyObject_Call(func, args, NULL); + Py_DECREF(args); + return result; +} +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallOneArg(PyObject *func, PyObject *arg) { +#if CYTHON_FAST_PYCALL + if (PyFunction_Check(func)) { + return __Pyx_PyFunction_FastCall(func, &arg, 1); + } +#endif + if (likely(PyCFunction_Check(func))) { + if (likely(PyCFunction_GET_FLAGS(func) & METH_O)) { + return __Pyx_PyObject_CallMethO(func, arg); +#if CYTHON_FAST_PYCCALL + } else if (PyCFunction_GET_FLAGS(func) & METH_FASTCALL) { + return __Pyx_PyCFunction_FastCall(func, &arg, 1); +#endif + } + } + return __Pyx__PyObject_CallOneArg(func, arg); +} +#else +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallOneArg(PyObject *func, PyObject *arg) { + PyObject *result; + PyObject *args = PyTuple_Pack(1, arg); + if (unlikely(!args)) return NULL; + result = __Pyx_PyObject_Call(func, args, NULL); + Py_DECREF(args); + return result; +} +#endif + +/* ObjectGetItem */ +#if CYTHON_USE_TYPE_SLOTS +static PyObject *__Pyx_PyObject_GetIndex(PyObject *obj, PyObject* index) { + PyObject *runerr; + Py_ssize_t key_value; + PySequenceMethods *m = Py_TYPE(obj)->tp_as_sequence; + if (unlikely(!(m && m->sq_item))) { + PyErr_Format(PyExc_TypeError, "'%.200s' object is not subscriptable", Py_TYPE(obj)->tp_name); + return NULL; + } + key_value = __Pyx_PyIndex_AsSsize_t(index); + if (likely(key_value != -1 || !(runerr = PyErr_Occurred()))) { + return __Pyx_GetItemInt_Fast(obj, key_value, 0, 1, 1); + } + if (PyErr_GivenExceptionMatches(runerr, PyExc_OverflowError)) { + PyErr_Clear(); + PyErr_Format(PyExc_IndexError, "cannot fit '%.200s' into an index-sized integer", Py_TYPE(index)->tp_name); + } + return NULL; +} +static PyObject *__Pyx_PyObject_GetItem(PyObject *obj, PyObject* key) { + PyMappingMethods *m = Py_TYPE(obj)->tp_as_mapping; + if (likely(m && m->mp_subscript)) { + return m->mp_subscript(obj, key); + } + return __Pyx_PyObject_GetIndex(obj, key); +} +#endif + +/* SliceTupleAndList */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE void __Pyx_crop_slice(Py_ssize_t* _start, Py_ssize_t* _stop, Py_ssize_t* _length) { + Py_ssize_t start = *_start, stop = *_stop, length = *_length; + if (start < 0) { + start += length; + if (start < 0) + start = 0; + } + if (stop < 0) + stop += length; + else if (stop > length) + stop = length; + *_length = stop - start; + *_start = start; + *_stop = stop; +} +static CYTHON_INLINE void __Pyx_copy_object_array(PyObject** CYTHON_RESTRICT src, PyObject** CYTHON_RESTRICT dest, Py_ssize_t length) { + PyObject *v; + Py_ssize_t i; + for (i = 0; i < length; i++) { + v = dest[i] = src[i]; + Py_INCREF(v); + } +} +static CYTHON_INLINE PyObject* __Pyx_PyList_GetSlice( + PyObject* src, Py_ssize_t start, Py_ssize_t stop) { + PyObject* dest; + Py_ssize_t length = PyList_GET_SIZE(src); + __Pyx_crop_slice(&start, &stop, &length); + if (unlikely(length <= 0)) + return PyList_New(0); + dest = PyList_New(length); + if (unlikely(!dest)) + return NULL; + __Pyx_copy_object_array( + ((PyListObject*)src)->ob_item + start, + ((PyListObject*)dest)->ob_item, + length); + return dest; +} +static CYTHON_INLINE PyObject* __Pyx_PyTuple_GetSlice( + PyObject* src, Py_ssize_t start, Py_ssize_t stop) { + PyObject* dest; + Py_ssize_t length = PyTuple_GET_SIZE(src); + __Pyx_crop_slice(&start, &stop, &length); + if (unlikely(length <= 0)) + return PyTuple_New(0); + dest = PyTuple_New(length); + if (unlikely(!dest)) + return NULL; + __Pyx_copy_object_array( + ((PyTupleObject*)src)->ob_item + start, + ((PyTupleObject*)dest)->ob_item, + length); + return dest; +} +#endif + +/* PyIntBinop */ +#if !CYTHON_COMPILING_IN_PYPY +static PyObject* __Pyx_PyInt_AddObjC(PyObject *op1, PyObject *op2, CYTHON_UNUSED long intval, int inplace, int zerodivision_check) { + (void)inplace; + (void)zerodivision_check; + #if PY_MAJOR_VERSION < 3 + if (likely(PyInt_CheckExact(op1))) { + const long b = intval; + long x; + long a = PyInt_AS_LONG(op1); + x = (long)((unsigned long)a + b); + if (likely((x^a) >= 0 || (x^b) >= 0)) + return PyInt_FromLong(x); + return PyLong_Type.tp_as_number->nb_add(op1, op2); + } + #endif + #if CYTHON_USE_PYLONG_INTERNALS + if (likely(PyLong_CheckExact(op1))) { + const long b = intval; + long a, x; +#ifdef HAVE_LONG_LONG + const PY_LONG_LONG llb = intval; + PY_LONG_LONG lla, llx; +#endif + const digit* digits = ((PyLongObject*)op1)->ob_digit; + const Py_ssize_t size = Py_SIZE(op1); + if (likely(__Pyx_sst_abs(size) <= 1)) { + a = likely(size) ? digits[0] : 0; + if (size == -1) a = -a; + } else { + switch (size) { + case -2: + if (8 * sizeof(long) - 1 > 2 * PyLong_SHIFT) { + a = -(long) (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0])); + break; +#ifdef HAVE_LONG_LONG + } else if (8 * sizeof(PY_LONG_LONG) - 1 > 2 * PyLong_SHIFT) { + lla = -(PY_LONG_LONG) (((((unsigned PY_LONG_LONG)digits[1]) << PyLong_SHIFT) | (unsigned PY_LONG_LONG)digits[0])); + goto long_long; +#endif + } + CYTHON_FALLTHROUGH; + case 2: + if (8 * sizeof(long) - 1 > 2 * PyLong_SHIFT) { + a = (long) (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0])); + break; +#ifdef HAVE_LONG_LONG + } else if (8 * sizeof(PY_LONG_LONG) - 1 > 2 * PyLong_SHIFT) { + lla = (PY_LONG_LONG) (((((unsigned PY_LONG_LONG)digits[1]) << PyLong_SHIFT) | (unsigned PY_LONG_LONG)digits[0])); + goto long_long; +#endif + } + CYTHON_FALLTHROUGH; + case -3: + if (8 * sizeof(long) - 1 > 3 * PyLong_SHIFT) { + a = -(long) (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0])); + break; +#ifdef HAVE_LONG_LONG + } else if (8 * sizeof(PY_LONG_LONG) - 1 > 3 * PyLong_SHIFT) { + lla = -(PY_LONG_LONG) (((((((unsigned PY_LONG_LONG)digits[2]) << PyLong_SHIFT) | (unsigned PY_LONG_LONG)digits[1]) << PyLong_SHIFT) | (unsigned PY_LONG_LONG)digits[0])); + goto long_long; +#endif + } + CYTHON_FALLTHROUGH; + case 3: + if (8 * sizeof(long) - 1 > 3 * PyLong_SHIFT) { + a = (long) (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0])); + break; +#ifdef HAVE_LONG_LONG + } else if (8 * sizeof(PY_LONG_LONG) - 1 > 3 * PyLong_SHIFT) { + lla = (PY_LONG_LONG) (((((((unsigned PY_LONG_LONG)digits[2]) << PyLong_SHIFT) | (unsigned PY_LONG_LONG)digits[1]) << PyLong_SHIFT) | (unsigned PY_LONG_LONG)digits[0])); + goto long_long; +#endif + } + CYTHON_FALLTHROUGH; + case -4: + if (8 * sizeof(long) - 1 > 4 * PyLong_SHIFT) { + a = -(long) (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0])); + break; +#ifdef HAVE_LONG_LONG + } else if (8 * sizeof(PY_LONG_LONG) - 1 > 4 * PyLong_SHIFT) { + lla = -(PY_LONG_LONG) (((((((((unsigned PY_LONG_LONG)digits[3]) << PyLong_SHIFT) | (unsigned PY_LONG_LONG)digits[2]) << PyLong_SHIFT) | (unsigned PY_LONG_LONG)digits[1]) << PyLong_SHIFT) | (unsigned PY_LONG_LONG)digits[0])); + goto long_long; +#endif + } + CYTHON_FALLTHROUGH; + case 4: + if (8 * sizeof(long) - 1 > 4 * PyLong_SHIFT) { + a = (long) (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0])); + break; +#ifdef HAVE_LONG_LONG + } else if (8 * sizeof(PY_LONG_LONG) - 1 > 4 * PyLong_SHIFT) { + lla = (PY_LONG_LONG) (((((((((unsigned PY_LONG_LONG)digits[3]) << PyLong_SHIFT) | (unsigned PY_LONG_LONG)digits[2]) << PyLong_SHIFT) | (unsigned PY_LONG_LONG)digits[1]) << PyLong_SHIFT) | (unsigned PY_LONG_LONG)digits[0])); + goto long_long; +#endif + } + CYTHON_FALLTHROUGH; + default: return PyLong_Type.tp_as_number->nb_add(op1, op2); + } + } + x = a + b; + return PyLong_FromLong(x); +#ifdef HAVE_LONG_LONG + long_long: + llx = lla + llb; + return PyLong_FromLongLong(llx); +#endif + + + } + #endif + if (PyFloat_CheckExact(op1)) { + const long b = intval; + double a = PyFloat_AS_DOUBLE(op1); + double result; + PyFPE_START_PROTECT("add", return NULL) + result = ((double)a) + (double)b; + PyFPE_END_PROTECT(result) + return PyFloat_FromDouble(result); + } + return (inplace ? PyNumber_InPlaceAdd : PyNumber_Add)(op1, op2); +} +#endif + +/* RaiseArgTupleInvalid */ +static void __Pyx_RaiseArgtupleInvalid( + const char* func_name, + int exact, + Py_ssize_t num_min, + Py_ssize_t num_max, + Py_ssize_t num_found) +{ + Py_ssize_t num_expected; + const char *more_or_less; + if (num_found < num_min) { + num_expected = num_min; + more_or_less = "at least"; + } else { + num_expected = num_max; + more_or_less = "at most"; + } + if (exact) { + more_or_less = "exactly"; + } + PyErr_Format(PyExc_TypeError, + "%.200s() takes %.8s %" CYTHON_FORMAT_SSIZE_T "d positional argument%.1s (%" CYTHON_FORMAT_SSIZE_T "d given)", + func_name, more_or_less, num_expected, + (num_expected == 1) ? "" : "s", num_found); +} + +/* RaiseDoubleKeywords */ +static void __Pyx_RaiseDoubleKeywordsError( + const char* func_name, + PyObject* kw_name) +{ + PyErr_Format(PyExc_TypeError, + #if PY_MAJOR_VERSION >= 3 + "%s() got multiple values for keyword argument '%U'", func_name, kw_name); + #else + "%s() got multiple values for keyword argument '%s'", func_name, + PyString_AsString(kw_name)); + #endif +} + +/* ParseKeywords */ +static int __Pyx_ParseOptionalKeywords( + PyObject *kwds, + PyObject **argnames[], + PyObject *kwds2, + PyObject *values[], + Py_ssize_t num_pos_args, + const char* function_name) +{ + PyObject *key = 0, *value = 0; + Py_ssize_t pos = 0; + PyObject*** name; + PyObject*** first_kw_arg = argnames + num_pos_args; + while (PyDict_Next(kwds, &pos, &key, &value)) { + name = first_kw_arg; + while (*name && (**name != key)) name++; + if (*name) { + values[name-argnames] = value; + continue; + } + name = first_kw_arg; + #if PY_MAJOR_VERSION < 3 + if (likely(PyString_CheckExact(key)) || likely(PyString_Check(key))) { + while (*name) { + if ((CYTHON_COMPILING_IN_PYPY || PyString_GET_SIZE(**name) == PyString_GET_SIZE(key)) + && _PyString_Eq(**name, key)) { + values[name-argnames] = value; + break; + } + name++; + } + if (*name) continue; + else { + PyObject*** argname = argnames; + while (argname != first_kw_arg) { + if ((**argname == key) || ( + (CYTHON_COMPILING_IN_PYPY || PyString_GET_SIZE(**argname) == PyString_GET_SIZE(key)) + && _PyString_Eq(**argname, key))) { + goto arg_passed_twice; + } + argname++; + } + } + } else + #endif + if (likely(PyUnicode_Check(key))) { + while (*name) { + int cmp = (**name == key) ? 0 : + #if !CYTHON_COMPILING_IN_PYPY && PY_MAJOR_VERSION >= 3 + (PyUnicode_GET_SIZE(**name) != PyUnicode_GET_SIZE(key)) ? 1 : + #endif + PyUnicode_Compare(**name, key); + if (cmp < 0 && unlikely(PyErr_Occurred())) goto bad; + if (cmp == 0) { + values[name-argnames] = value; + break; + } + name++; + } + if (*name) continue; + else { + PyObject*** argname = argnames; + while (argname != first_kw_arg) { + int cmp = (**argname == key) ? 0 : + #if !CYTHON_COMPILING_IN_PYPY && PY_MAJOR_VERSION >= 3 + (PyUnicode_GET_SIZE(**argname) != PyUnicode_GET_SIZE(key)) ? 1 : + #endif + PyUnicode_Compare(**argname, key); + if (cmp < 0 && unlikely(PyErr_Occurred())) goto bad; + if (cmp == 0) goto arg_passed_twice; + argname++; + } + } + } else + goto invalid_keyword_type; + if (kwds2) { + if (unlikely(PyDict_SetItem(kwds2, key, value))) goto bad; + } else { + goto invalid_keyword; + } + } + return 0; +arg_passed_twice: + __Pyx_RaiseDoubleKeywordsError(function_name, key); + goto bad; +invalid_keyword_type: + PyErr_Format(PyExc_TypeError, + "%.200s() keywords must be strings", function_name); + goto bad; +invalid_keyword: + PyErr_Format(PyExc_TypeError, + #if PY_MAJOR_VERSION < 3 + "%.200s() got an unexpected keyword argument '%.200s'", + function_name, PyString_AsString(key)); + #else + "%s() got an unexpected keyword argument '%U'", + function_name, key); + #endif +bad: + return -1; +} + +/* RaiseTooManyValuesToUnpack */ +static CYTHON_INLINE void __Pyx_RaiseTooManyValuesError(Py_ssize_t expected) { + PyErr_Format(PyExc_ValueError, + "too many values to unpack (expected %" CYTHON_FORMAT_SSIZE_T "d)", expected); +} + +/* RaiseNeedMoreValuesToUnpack */ +static CYTHON_INLINE void __Pyx_RaiseNeedMoreValuesError(Py_ssize_t index) { + PyErr_Format(PyExc_ValueError, + "need more than %" CYTHON_FORMAT_SSIZE_T "d value%.1s to unpack", + index, (index == 1) ? "" : "s"); +} + +/* IterFinish */ +static CYTHON_INLINE int __Pyx_IterFinish(void) { +#if CYTHON_FAST_THREAD_STATE + PyThreadState *tstate = __Pyx_PyThreadState_Current; + PyObject* exc_type = tstate->curexc_type; + if (unlikely(exc_type)) { + if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) { + PyObject *exc_value, *exc_tb; + exc_value = tstate->curexc_value; + exc_tb = tstate->curexc_traceback; + tstate->curexc_type = 0; + tstate->curexc_value = 0; + tstate->curexc_traceback = 0; + Py_DECREF(exc_type); + Py_XDECREF(exc_value); + Py_XDECREF(exc_tb); + return 0; + } else { + return -1; + } + } + return 0; +#else + if (unlikely(PyErr_Occurred())) { + if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) { + PyErr_Clear(); + return 0; + } else { + return -1; + } + } + return 0; +#endif +} + +/* UnpackItemEndCheck */ +static int __Pyx_IternextUnpackEndCheck(PyObject *retval, Py_ssize_t expected) { + if (unlikely(retval)) { + Py_DECREF(retval); + __Pyx_RaiseTooManyValuesError(expected); + return -1; + } else { + return __Pyx_IterFinish(); + } + return 0; +} + +/* PyErrFetchRestore */ +#if CYTHON_FAST_THREAD_STATE +static CYTHON_INLINE void __Pyx_ErrRestoreInState(PyThreadState *tstate, PyObject *type, PyObject *value, PyObject *tb) { + PyObject *tmp_type, *tmp_value, *tmp_tb; + tmp_type = tstate->curexc_type; + tmp_value = tstate->curexc_value; + tmp_tb = tstate->curexc_traceback; + tstate->curexc_type = type; + tstate->curexc_value = value; + tstate->curexc_traceback = tb; + Py_XDECREF(tmp_type); + Py_XDECREF(tmp_value); + Py_XDECREF(tmp_tb); +} +static CYTHON_INLINE void __Pyx_ErrFetchInState(PyThreadState *tstate, PyObject **type, PyObject **value, PyObject **tb) { + *type = tstate->curexc_type; + *value = tstate->curexc_value; + *tb = tstate->curexc_traceback; + tstate->curexc_type = 0; + tstate->curexc_value = 0; + tstate->curexc_traceback = 0; +} +#endif + +/* RaiseException */ +#if PY_MAJOR_VERSION < 3 +static void __Pyx_Raise(PyObject *type, PyObject *value, PyObject *tb, + CYTHON_UNUSED PyObject *cause) { + __Pyx_PyThreadState_declare + Py_XINCREF(type); + if (!value || value == Py_None) + value = NULL; + else + Py_INCREF(value); + if (!tb || tb == Py_None) + tb = NULL; + else { + Py_INCREF(tb); + if (!PyTraceBack_Check(tb)) { + PyErr_SetString(PyExc_TypeError, + "raise: arg 3 must be a traceback or None"); + goto raise_error; + } + } + if (PyType_Check(type)) { +#if CYTHON_COMPILING_IN_PYPY + if (!value) { + Py_INCREF(Py_None); + value = Py_None; + } +#endif + PyErr_NormalizeException(&type, &value, &tb); + } else { + if (value) { + PyErr_SetString(PyExc_TypeError, + "instance exception may not have a separate value"); + goto raise_error; + } + value = type; + type = (PyObject*) Py_TYPE(type); + Py_INCREF(type); + if (!PyType_IsSubtype((PyTypeObject *)type, (PyTypeObject *)PyExc_BaseException)) { + PyErr_SetString(PyExc_TypeError, + "raise: exception class must be a subclass of BaseException"); + goto raise_error; + } + } + __Pyx_PyThreadState_assign + __Pyx_ErrRestore(type, value, tb); + return; +raise_error: + Py_XDECREF(value); + Py_XDECREF(type); + Py_XDECREF(tb); + return; +} +#else +static void __Pyx_Raise(PyObject *type, PyObject *value, PyObject *tb, PyObject *cause) { + PyObject* owned_instance = NULL; + if (tb == Py_None) { + tb = 0; + } else if (tb && !PyTraceBack_Check(tb)) { + PyErr_SetString(PyExc_TypeError, + "raise: arg 3 must be a traceback or None"); + goto bad; + } + if (value == Py_None) + value = 0; + if (PyExceptionInstance_Check(type)) { + if (value) { + PyErr_SetString(PyExc_TypeError, + "instance exception may not have a separate value"); + goto bad; + } + value = type; + type = (PyObject*) Py_TYPE(value); + } else if (PyExceptionClass_Check(type)) { + PyObject *instance_class = NULL; + if (value && PyExceptionInstance_Check(value)) { + instance_class = (PyObject*) Py_TYPE(value); + if (instance_class != type) { + int is_subclass = PyObject_IsSubclass(instance_class, type); + if (!is_subclass) { + instance_class = NULL; + } else if (unlikely(is_subclass == -1)) { + goto bad; + } else { + type = instance_class; + } + } + } + if (!instance_class) { + PyObject *args; + if (!value) + args = PyTuple_New(0); + else if (PyTuple_Check(value)) { + Py_INCREF(value); + args = value; + } else + args = PyTuple_Pack(1, value); + if (!args) + goto bad; + owned_instance = PyObject_Call(type, args, NULL); + Py_DECREF(args); + if (!owned_instance) + goto bad; + value = owned_instance; + if (!PyExceptionInstance_Check(value)) { + PyErr_Format(PyExc_TypeError, + "calling %R should have returned an instance of " + "BaseException, not %R", + type, Py_TYPE(value)); + goto bad; + } + } + } else { + PyErr_SetString(PyExc_TypeError, + "raise: exception class must be a subclass of BaseException"); + goto bad; + } + if (cause) { + PyObject *fixed_cause; + if (cause == Py_None) { + fixed_cause = NULL; + } else if (PyExceptionClass_Check(cause)) { + fixed_cause = PyObject_CallObject(cause, NULL); + if (fixed_cause == NULL) + goto bad; + } else if (PyExceptionInstance_Check(cause)) { + fixed_cause = cause; + Py_INCREF(fixed_cause); + } else { + PyErr_SetString(PyExc_TypeError, + "exception causes must derive from " + "BaseException"); + goto bad; + } + PyException_SetCause(value, fixed_cause); + } + PyErr_SetObject(type, value); + if (tb) { +#if CYTHON_COMPILING_IN_PYPY + PyObject *tmp_type, *tmp_value, *tmp_tb; + PyErr_Fetch(&tmp_type, &tmp_value, &tmp_tb); + Py_INCREF(tb); + PyErr_Restore(tmp_type, tmp_value, tb); + Py_XDECREF(tmp_tb); +#else + PyThreadState *tstate = __Pyx_PyThreadState_Current; + PyObject* tmp_tb = tstate->curexc_traceback; + if (tb != tmp_tb) { + Py_INCREF(tb); + tstate->curexc_traceback = tb; + Py_XDECREF(tmp_tb); + } +#endif + } +bad: + Py_XDECREF(owned_instance); + return; +} +#endif + +/* DictGetItem */ +#if PY_MAJOR_VERSION >= 3 && !CYTHON_COMPILING_IN_PYPY +static PyObject *__Pyx_PyDict_GetItem(PyObject *d, PyObject* key) { + PyObject *value; + value = PyDict_GetItemWithError(d, key); + if (unlikely(!value)) { + if (!PyErr_Occurred()) { + if (unlikely(PyTuple_Check(key))) { + PyObject* args = PyTuple_Pack(1, key); + if (likely(args)) { + PyErr_SetObject(PyExc_KeyError, args); + Py_DECREF(args); + } + } else { + PyErr_SetObject(PyExc_KeyError, key); + } + } + return NULL; + } + Py_INCREF(value); + return value; +} +#endif + +/* RaiseNoneIterError */ +static CYTHON_INLINE void __Pyx_RaiseNoneNotIterableError(void) { + PyErr_SetString(PyExc_TypeError, "'NoneType' object is not iterable"); +} + +/* GetTopmostException */ +#if CYTHON_USE_EXC_INFO_STACK +static _PyErr_StackItem * +__Pyx_PyErr_GetTopmostException(PyThreadState *tstate) +{ + _PyErr_StackItem *exc_info = tstate->exc_info; + while ((exc_info->exc_type == NULL || exc_info->exc_type == Py_None) && + exc_info->previous_item != NULL) + { + exc_info = exc_info->previous_item; + } + return exc_info; +} +#endif + +/* SaveResetException */ +#if CYTHON_FAST_THREAD_STATE +static CYTHON_INLINE void __Pyx__ExceptionSave(PyThreadState *tstate, PyObject **type, PyObject **value, PyObject **tb) { + #if CYTHON_USE_EXC_INFO_STACK + _PyErr_StackItem *exc_info = __Pyx_PyErr_GetTopmostException(tstate); + *type = exc_info->exc_type; + *value = exc_info->exc_value; + *tb = exc_info->exc_traceback; + #else + *type = tstate->exc_type; + *value = tstate->exc_value; + *tb = tstate->exc_traceback; + #endif + Py_XINCREF(*type); + Py_XINCREF(*value); + Py_XINCREF(*tb); +} +static CYTHON_INLINE void __Pyx__ExceptionReset(PyThreadState *tstate, PyObject *type, PyObject *value, PyObject *tb) { + PyObject *tmp_type, *tmp_value, *tmp_tb; + #if CYTHON_USE_EXC_INFO_STACK + _PyErr_StackItem *exc_info = tstate->exc_info; + tmp_type = exc_info->exc_type; + tmp_value = exc_info->exc_value; + tmp_tb = exc_info->exc_traceback; + exc_info->exc_type = type; + exc_info->exc_value = value; + exc_info->exc_traceback = tb; + #else + tmp_type = tstate->exc_type; + tmp_value = tstate->exc_value; + tmp_tb = tstate->exc_traceback; + tstate->exc_type = type; + tstate->exc_value = value; + tstate->exc_traceback = tb; + #endif + Py_XDECREF(tmp_type); + Py_XDECREF(tmp_value); + Py_XDECREF(tmp_tb); +} +#endif + +/* PyErrExceptionMatches */ +#if CYTHON_FAST_THREAD_STATE +static int __Pyx_PyErr_ExceptionMatchesTuple(PyObject *exc_type, PyObject *tuple) { + Py_ssize_t i, n; + n = PyTuple_GET_SIZE(tuple); +#if PY_MAJOR_VERSION >= 3 + for (i=0; icurexc_type; + if (exc_type == err) return 1; + if (unlikely(!exc_type)) return 0; + if (unlikely(PyTuple_Check(err))) + return __Pyx_PyErr_ExceptionMatchesTuple(exc_type, err); + return __Pyx_PyErr_GivenExceptionMatches(exc_type, err); +} +#endif + +/* GetException */ +#if CYTHON_FAST_THREAD_STATE +static int __Pyx__GetException(PyThreadState *tstate, PyObject **type, PyObject **value, PyObject **tb) +#else +static int __Pyx_GetException(PyObject **type, PyObject **value, PyObject **tb) +#endif +{ + PyObject *local_type, *local_value, *local_tb; +#if CYTHON_FAST_THREAD_STATE + PyObject *tmp_type, *tmp_value, *tmp_tb; + local_type = tstate->curexc_type; + local_value = tstate->curexc_value; + local_tb = tstate->curexc_traceback; + tstate->curexc_type = 0; + tstate->curexc_value = 0; + tstate->curexc_traceback = 0; +#else + PyErr_Fetch(&local_type, &local_value, &local_tb); +#endif + PyErr_NormalizeException(&local_type, &local_value, &local_tb); +#if CYTHON_FAST_THREAD_STATE + if (unlikely(tstate->curexc_type)) +#else + if (unlikely(PyErr_Occurred())) +#endif + goto bad; + #if PY_MAJOR_VERSION >= 3 + if (local_tb) { + if (unlikely(PyException_SetTraceback(local_value, local_tb) < 0)) + goto bad; + } + #endif + Py_XINCREF(local_tb); + Py_XINCREF(local_type); + Py_XINCREF(local_value); + *type = local_type; + *value = local_value; + *tb = local_tb; +#if CYTHON_FAST_THREAD_STATE + #if CYTHON_USE_EXC_INFO_STACK + { + _PyErr_StackItem *exc_info = tstate->exc_info; + tmp_type = exc_info->exc_type; + tmp_value = exc_info->exc_value; + tmp_tb = exc_info->exc_traceback; + exc_info->exc_type = local_type; + exc_info->exc_value = local_value; + exc_info->exc_traceback = local_tb; + } + #else + tmp_type = tstate->exc_type; + tmp_value = tstate->exc_value; + tmp_tb = tstate->exc_traceback; + tstate->exc_type = local_type; + tstate->exc_value = local_value; + tstate->exc_traceback = local_tb; + #endif + Py_XDECREF(tmp_type); + Py_XDECREF(tmp_value); + Py_XDECREF(tmp_tb); +#else + PyErr_SetExcInfo(local_type, local_value, local_tb); +#endif + return 0; +bad: + *type = 0; + *value = 0; + *tb = 0; + Py_XDECREF(local_type); + Py_XDECREF(local_value); + Py_XDECREF(local_tb); + return -1; +} + +/* TypeImport */ +#ifndef __PYX_HAVE_RT_ImportType +#define __PYX_HAVE_RT_ImportType +static PyTypeObject *__Pyx_ImportType(PyObject *module, const char *module_name, const char *class_name, + size_t size, enum __Pyx_ImportType_CheckSize check_size) +{ + PyObject *result = 0; + char warning[200]; + Py_ssize_t basicsize; +#ifdef Py_LIMITED_API + PyObject *py_basicsize; +#endif + result = PyObject_GetAttrString(module, class_name); + if (!result) + goto bad; + if (!PyType_Check(result)) { + PyErr_Format(PyExc_TypeError, + "%.200s.%.200s is not a type object", + module_name, class_name); + goto bad; + } +#ifndef Py_LIMITED_API + basicsize = ((PyTypeObject *)result)->tp_basicsize; +#else + py_basicsize = PyObject_GetAttrString(result, "__basicsize__"); + if (!py_basicsize) + goto bad; + basicsize = PyLong_AsSsize_t(py_basicsize); + Py_DECREF(py_basicsize); + py_basicsize = 0; + if (basicsize == (Py_ssize_t)-1 && PyErr_Occurred()) + goto bad; +#endif + if ((size_t)basicsize < size) { + PyErr_Format(PyExc_ValueError, + "%.200s.%.200s size changed, may indicate binary incompatibility. " + "Expected %zd from C header, got %zd from PyObject", + module_name, class_name, size, basicsize); + goto bad; + } + if (check_size == __Pyx_ImportType_CheckSize_Error && (size_t)basicsize != size) { + PyErr_Format(PyExc_ValueError, + "%.200s.%.200s size changed, may indicate binary incompatibility. " + "Expected %zd from C header, got %zd from PyObject", + module_name, class_name, size, basicsize); + goto bad; + } + else if (check_size == __Pyx_ImportType_CheckSize_Warn && (size_t)basicsize > size) { + PyOS_snprintf(warning, sizeof(warning), + "%s.%s size changed, may indicate binary incompatibility. " + "Expected %zd from C header, got %zd from PyObject", + module_name, class_name, size, basicsize); + if (PyErr_WarnEx(NULL, warning, 0) < 0) goto bad; + } + return (PyTypeObject *)result; +bad: + Py_XDECREF(result); + return NULL; +} +#endif + +/* Import */ +static PyObject *__Pyx_Import(PyObject *name, PyObject *from_list, int level) { + PyObject *empty_list = 0; + PyObject *module = 0; + PyObject *global_dict = 0; + PyObject *empty_dict = 0; + PyObject *list; + #if PY_MAJOR_VERSION < 3 + PyObject *py_import; + py_import = __Pyx_PyObject_GetAttrStr(__pyx_b, __pyx_n_s_import); + if (!py_import) + goto bad; + #endif + if (from_list) + list = from_list; + else { + empty_list = PyList_New(0); + if (!empty_list) + goto bad; + list = empty_list; + } + global_dict = PyModule_GetDict(__pyx_m); + if (!global_dict) + goto bad; + empty_dict = PyDict_New(); + if (!empty_dict) + goto bad; + { + #if PY_MAJOR_VERSION >= 3 + if (level == -1) { + if (strchr(__Pyx_MODULE_NAME, '.')) { + module = PyImport_ImportModuleLevelObject( + name, global_dict, empty_dict, list, 1); + if (!module) { + if (!PyErr_ExceptionMatches(PyExc_ImportError)) + goto bad; + PyErr_Clear(); + } + } + level = 0; + } + #endif + if (!module) { + #if PY_MAJOR_VERSION < 3 + PyObject *py_level = PyInt_FromLong(level); + if (!py_level) + goto bad; + module = PyObject_CallFunctionObjArgs(py_import, + name, global_dict, empty_dict, list, py_level, (PyObject *)NULL); + Py_DECREF(py_level); + #else + module = PyImport_ImportModuleLevelObject( + name, global_dict, empty_dict, list, level); + #endif + } + } +bad: + #if PY_MAJOR_VERSION < 3 + Py_XDECREF(py_import); + #endif + Py_XDECREF(empty_list); + Py_XDECREF(empty_dict); + return module; +} + +/* ImportFrom */ +static PyObject* __Pyx_ImportFrom(PyObject* module, PyObject* name) { + PyObject* value = __Pyx_PyObject_GetAttrStr(module, name); + if (unlikely(!value) && PyErr_ExceptionMatches(PyExc_AttributeError)) { + PyErr_Format(PyExc_ImportError, + #if PY_MAJOR_VERSION < 3 + "cannot import name %.230s", PyString_AS_STRING(name)); + #else + "cannot import name %S", name); + #endif + } + return value; +} + +/* PyObjectCallNoArg */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallNoArg(PyObject *func) { +#if CYTHON_FAST_PYCALL + if (PyFunction_Check(func)) { + return __Pyx_PyFunction_FastCall(func, NULL, 0); + } +#endif +#ifdef __Pyx_CyFunction_USED + if (likely(PyCFunction_Check(func) || __Pyx_CyFunction_Check(func))) +#else + if (likely(PyCFunction_Check(func))) +#endif + { + if (likely(PyCFunction_GET_FLAGS(func) & METH_NOARGS)) { + return __Pyx_PyObject_CallMethO(func, NULL); + } + } + return __Pyx_PyObject_Call(func, __pyx_empty_tuple, NULL); +} +#endif + +/* CLineInTraceback */ +#ifndef CYTHON_CLINE_IN_TRACEBACK +static int __Pyx_CLineForTraceback(PyThreadState *tstate, int c_line) { + PyObject *use_cline; + PyObject *ptype, *pvalue, *ptraceback; +#if CYTHON_COMPILING_IN_CPYTHON + PyObject **cython_runtime_dict; +#endif + if (unlikely(!__pyx_cython_runtime)) { + return c_line; + } + __Pyx_ErrFetchInState(tstate, &ptype, &pvalue, &ptraceback); +#if CYTHON_COMPILING_IN_CPYTHON + cython_runtime_dict = _PyObject_GetDictPtr(__pyx_cython_runtime); + if (likely(cython_runtime_dict)) { + __PYX_PY_DICT_LOOKUP_IF_MODIFIED( + use_cline, *cython_runtime_dict, + __Pyx_PyDict_GetItemStr(*cython_runtime_dict, __pyx_n_s_cline_in_traceback)) + } else +#endif + { + PyObject *use_cline_obj = __Pyx_PyObject_GetAttrStr(__pyx_cython_runtime, __pyx_n_s_cline_in_traceback); + if (use_cline_obj) { + use_cline = PyObject_Not(use_cline_obj) ? Py_False : Py_True; + Py_DECREF(use_cline_obj); + } else { + PyErr_Clear(); + use_cline = NULL; + } + } + if (!use_cline) { + c_line = 0; + PyObject_SetAttr(__pyx_cython_runtime, __pyx_n_s_cline_in_traceback, Py_False); + } + else if (use_cline == Py_False || (use_cline != Py_True && PyObject_Not(use_cline) != 0)) { + c_line = 0; + } + __Pyx_ErrRestoreInState(tstate, ptype, pvalue, ptraceback); + return c_line; +} +#endif + +/* CodeObjectCache */ +static int __pyx_bisect_code_objects(__Pyx_CodeObjectCacheEntry* entries, int count, int code_line) { + int start = 0, mid = 0, end = count - 1; + if (end >= 0 && code_line > entries[end].code_line) { + return count; + } + while (start < end) { + mid = start + (end - start) / 2; + if (code_line < entries[mid].code_line) { + end = mid; + } else if (code_line > entries[mid].code_line) { + start = mid + 1; + } else { + return mid; + } + } + if (code_line <= entries[mid].code_line) { + return mid; + } else { + return mid + 1; + } +} +static PyCodeObject *__pyx_find_code_object(int code_line) { + PyCodeObject* code_object; + int pos; + if (unlikely(!code_line) || unlikely(!__pyx_code_cache.entries)) { + return NULL; + } + pos = __pyx_bisect_code_objects(__pyx_code_cache.entries, __pyx_code_cache.count, code_line); + if (unlikely(pos >= __pyx_code_cache.count) || unlikely(__pyx_code_cache.entries[pos].code_line != code_line)) { + return NULL; + } + code_object = __pyx_code_cache.entries[pos].code_object; + Py_INCREF(code_object); + return code_object; +} +static void __pyx_insert_code_object(int code_line, PyCodeObject* code_object) { + int pos, i; + __Pyx_CodeObjectCacheEntry* entries = __pyx_code_cache.entries; + if (unlikely(!code_line)) { + return; + } + if (unlikely(!entries)) { + entries = (__Pyx_CodeObjectCacheEntry*)PyMem_Malloc(64*sizeof(__Pyx_CodeObjectCacheEntry)); + if (likely(entries)) { + __pyx_code_cache.entries = entries; + __pyx_code_cache.max_count = 64; + __pyx_code_cache.count = 1; + entries[0].code_line = code_line; + entries[0].code_object = code_object; + Py_INCREF(code_object); + } + return; + } + pos = __pyx_bisect_code_objects(__pyx_code_cache.entries, __pyx_code_cache.count, code_line); + if ((pos < __pyx_code_cache.count) && unlikely(__pyx_code_cache.entries[pos].code_line == code_line)) { + PyCodeObject* tmp = entries[pos].code_object; + entries[pos].code_object = code_object; + Py_DECREF(tmp); + return; + } + if (__pyx_code_cache.count == __pyx_code_cache.max_count) { + int new_max = __pyx_code_cache.max_count + 64; + entries = (__Pyx_CodeObjectCacheEntry*)PyMem_Realloc( + __pyx_code_cache.entries, (size_t)new_max*sizeof(__Pyx_CodeObjectCacheEntry)); + if (unlikely(!entries)) { + return; + } + __pyx_code_cache.entries = entries; + __pyx_code_cache.max_count = new_max; + } + for (i=__pyx_code_cache.count; i>pos; i--) { + entries[i] = entries[i-1]; + } + entries[pos].code_line = code_line; + entries[pos].code_object = code_object; + __pyx_code_cache.count++; + Py_INCREF(code_object); +} + +/* AddTraceback */ +#include "compile.h" +#include "frameobject.h" +#include "traceback.h" +static PyCodeObject* __Pyx_CreateCodeObjectForTraceback( + const char *funcname, int c_line, + int py_line, const char *filename) { + PyCodeObject *py_code = 0; + PyObject *py_srcfile = 0; + PyObject *py_funcname = 0; + #if PY_MAJOR_VERSION < 3 + py_srcfile = PyString_FromString(filename); + #else + py_srcfile = PyUnicode_FromString(filename); + #endif + if (!py_srcfile) goto bad; + if (c_line) { + #if PY_MAJOR_VERSION < 3 + py_funcname = PyString_FromFormat( "%s (%s:%d)", funcname, __pyx_cfilenm, c_line); + #else + py_funcname = PyUnicode_FromFormat( "%s (%s:%d)", funcname, __pyx_cfilenm, c_line); + #endif + } + else { + #if PY_MAJOR_VERSION < 3 + py_funcname = PyString_FromString(funcname); + #else + py_funcname = PyUnicode_FromString(funcname); + #endif + } + if (!py_funcname) goto bad; + py_code = __Pyx_PyCode_New( + 0, + 0, + 0, + 0, + 0, + __pyx_empty_bytes, /*PyObject *code,*/ + __pyx_empty_tuple, /*PyObject *consts,*/ + __pyx_empty_tuple, /*PyObject *names,*/ + __pyx_empty_tuple, /*PyObject *varnames,*/ + __pyx_empty_tuple, /*PyObject *freevars,*/ + __pyx_empty_tuple, /*PyObject *cellvars,*/ + py_srcfile, /*PyObject *filename,*/ + py_funcname, /*PyObject *name,*/ + py_line, + __pyx_empty_bytes /*PyObject *lnotab*/ + ); + Py_DECREF(py_srcfile); + Py_DECREF(py_funcname); + return py_code; +bad: + Py_XDECREF(py_srcfile); + Py_XDECREF(py_funcname); + return NULL; +} +static void __Pyx_AddTraceback(const char *funcname, int c_line, + int py_line, const char *filename) { + PyCodeObject *py_code = 0; + PyFrameObject *py_frame = 0; + PyThreadState *tstate = __Pyx_PyThreadState_Current; + if (c_line) { + c_line = __Pyx_CLineForTraceback(tstate, c_line); + } + py_code = __pyx_find_code_object(c_line ? -c_line : py_line); + if (!py_code) { + py_code = __Pyx_CreateCodeObjectForTraceback( + funcname, c_line, py_line, filename); + if (!py_code) goto bad; + __pyx_insert_code_object(c_line ? -c_line : py_line, py_code); + } + py_frame = PyFrame_New( + tstate, /*PyThreadState *tstate,*/ + py_code, /*PyCodeObject *code,*/ + __pyx_d, /*PyObject *globals,*/ + 0 /*PyObject *locals*/ + ); + if (!py_frame) goto bad; + __Pyx_PyFrame_SetLineNumber(py_frame, py_line); + PyTraceBack_Here(py_frame); +bad: + Py_XDECREF(py_code); + Py_XDECREF(py_frame); +} + +/* CIntFromPyVerify */ +#define __PYX_VERIFY_RETURN_INT(target_type, func_type, func_value)\ + __PYX__VERIFY_RETURN_INT(target_type, func_type, func_value, 0) +#define __PYX_VERIFY_RETURN_INT_EXC(target_type, func_type, func_value)\ + __PYX__VERIFY_RETURN_INT(target_type, func_type, func_value, 1) +#define __PYX__VERIFY_RETURN_INT(target_type, func_type, func_value, exc)\ + {\ + func_type value = func_value;\ + if (sizeof(target_type) < sizeof(func_type)) {\ + if (unlikely(value != (func_type) (target_type) value)) {\ + func_type zero = 0;\ + if (exc && unlikely(value == (func_type)-1 && PyErr_Occurred()))\ + return (target_type) -1;\ + if (is_unsigned && unlikely(value < zero))\ + goto raise_neg_overflow;\ + else\ + goto raise_overflow;\ + }\ + }\ + return (target_type) value;\ + } + +/* CIntToPy */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_long(long value) { + const long neg_one = (long) ((long) 0 - (long) 1), const_zero = (long) 0; + const int is_unsigned = neg_one > const_zero; + if (is_unsigned) { + if (sizeof(long) < sizeof(long)) { + return PyInt_FromLong((long) value); + } else if (sizeof(long) <= sizeof(unsigned long)) { + return PyLong_FromUnsignedLong((unsigned long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(long) <= sizeof(unsigned PY_LONG_LONG)) { + return PyLong_FromUnsignedLongLong((unsigned PY_LONG_LONG) value); +#endif + } + } else { + if (sizeof(long) <= sizeof(long)) { + return PyInt_FromLong((long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(long) <= sizeof(PY_LONG_LONG)) { + return PyLong_FromLongLong((PY_LONG_LONG) value); +#endif + } + } + { + int one = 1; int little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&value; + return _PyLong_FromByteArray(bytes, sizeof(long), + little, !is_unsigned); + } +} + +/* CIntToPy */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_int(int value) { + const int neg_one = (int) ((int) 0 - (int) 1), const_zero = (int) 0; + const int is_unsigned = neg_one > const_zero; + if (is_unsigned) { + if (sizeof(int) < sizeof(long)) { + return PyInt_FromLong((long) value); + } else if (sizeof(int) <= sizeof(unsigned long)) { + return PyLong_FromUnsignedLong((unsigned long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(int) <= sizeof(unsigned PY_LONG_LONG)) { + return PyLong_FromUnsignedLongLong((unsigned PY_LONG_LONG) value); +#endif + } + } else { + if (sizeof(int) <= sizeof(long)) { + return PyInt_FromLong((long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(int) <= sizeof(PY_LONG_LONG)) { + return PyLong_FromLongLong((PY_LONG_LONG) value); +#endif + } + } + { + int one = 1; int little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&value; + return _PyLong_FromByteArray(bytes, sizeof(int), + little, !is_unsigned); + } +} + +/* CIntToPy */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_npy_uint32(npy_uint32 value) { + const npy_uint32 neg_one = (npy_uint32) ((npy_uint32) 0 - (npy_uint32) 1), const_zero = (npy_uint32) 0; + const int is_unsigned = neg_one > const_zero; + if (is_unsigned) { + if (sizeof(npy_uint32) < sizeof(long)) { + return PyInt_FromLong((long) value); + } else if (sizeof(npy_uint32) <= sizeof(unsigned long)) { + return PyLong_FromUnsignedLong((unsigned long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(npy_uint32) <= sizeof(unsigned PY_LONG_LONG)) { + return PyLong_FromUnsignedLongLong((unsigned PY_LONG_LONG) value); +#endif + } + } else { + if (sizeof(npy_uint32) <= sizeof(long)) { + return PyInt_FromLong((long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(npy_uint32) <= sizeof(PY_LONG_LONG)) { + return PyLong_FromLongLong((PY_LONG_LONG) value); +#endif + } + } + { + int one = 1; int little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&value; + return _PyLong_FromByteArray(bytes, sizeof(npy_uint32), + little, !is_unsigned); + } +} + +/* Declarations */ +#if CYTHON_CCOMPLEX + #ifdef __cplusplus + static CYTHON_INLINE __pyx_t_float_complex __pyx_t_float_complex_from_parts(float x, float y) { + return ::std::complex< float >(x, y); + } + #else + static CYTHON_INLINE __pyx_t_float_complex __pyx_t_float_complex_from_parts(float x, float y) { + return x + y*(__pyx_t_float_complex)_Complex_I; + } + #endif +#else + static CYTHON_INLINE __pyx_t_float_complex __pyx_t_float_complex_from_parts(float x, float y) { + __pyx_t_float_complex z; + z.real = x; + z.imag = y; + return z; + } +#endif + +/* Arithmetic */ +#if CYTHON_CCOMPLEX +#else + static CYTHON_INLINE int __Pyx_c_eq_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + return (a.real == b.real) && (a.imag == b.imag); + } + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_sum_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + __pyx_t_float_complex z; + z.real = a.real + b.real; + z.imag = a.imag + b.imag; + return z; + } + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_diff_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + __pyx_t_float_complex z; + z.real = a.real - b.real; + z.imag = a.imag - b.imag; + return z; + } + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_prod_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + __pyx_t_float_complex z; + z.real = a.real * b.real - a.imag * b.imag; + z.imag = a.real * b.imag + a.imag * b.real; + return z; + } + #if 1 + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_quot_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + if (b.imag == 0) { + return __pyx_t_float_complex_from_parts(a.real / b.real, a.imag / b.real); + } else if (fabsf(b.real) >= fabsf(b.imag)) { + if (b.real == 0 && b.imag == 0) { + return __pyx_t_float_complex_from_parts(a.real / b.real, a.imag / b.imag); + } else { + float r = b.imag / b.real; + float s = (float)(1.0) / (b.real + b.imag * r); + return __pyx_t_float_complex_from_parts( + (a.real + a.imag * r) * s, (a.imag - a.real * r) * s); + } + } else { + float r = b.real / b.imag; + float s = (float)(1.0) / (b.imag + b.real * r); + return __pyx_t_float_complex_from_parts( + (a.real * r + a.imag) * s, (a.imag * r - a.real) * s); + } + } + #else + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_quot_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + if (b.imag == 0) { + return __pyx_t_float_complex_from_parts(a.real / b.real, a.imag / b.real); + } else { + float denom = b.real * b.real + b.imag * b.imag; + return __pyx_t_float_complex_from_parts( + (a.real * b.real + a.imag * b.imag) / denom, + (a.imag * b.real - a.real * b.imag) / denom); + } + } + #endif + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_neg_float(__pyx_t_float_complex a) { + __pyx_t_float_complex z; + z.real = -a.real; + z.imag = -a.imag; + return z; + } + static CYTHON_INLINE int __Pyx_c_is_zero_float(__pyx_t_float_complex a) { + return (a.real == 0) && (a.imag == 0); + } + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_conj_float(__pyx_t_float_complex a) { + __pyx_t_float_complex z; + z.real = a.real; + z.imag = -a.imag; + return z; + } + #if 1 + static CYTHON_INLINE float __Pyx_c_abs_float(__pyx_t_float_complex z) { + #if !defined(HAVE_HYPOT) || defined(_MSC_VER) + return sqrtf(z.real*z.real + z.imag*z.imag); + #else + return hypotf(z.real, z.imag); + #endif + } + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_pow_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + __pyx_t_float_complex z; + float r, lnr, theta, z_r, z_theta; + if (b.imag == 0 && b.real == (int)b.real) { + if (b.real < 0) { + float denom = a.real * a.real + a.imag * a.imag; + a.real = a.real / denom; + a.imag = -a.imag / denom; + b.real = -b.real; + } + switch ((int)b.real) { + case 0: + z.real = 1; + z.imag = 0; + return z; + case 1: + return a; + case 2: + return __Pyx_c_prod_float(a, a); + case 3: + z = __Pyx_c_prod_float(a, a); + return __Pyx_c_prod_float(z, a); + case 4: + z = __Pyx_c_prod_float(a, a); + return __Pyx_c_prod_float(z, z); + } + } + if (a.imag == 0) { + if (a.real == 0) { + return a; + } else if (b.imag == 0) { + z.real = powf(a.real, b.real); + z.imag = 0; + return z; + } else if (a.real > 0) { + r = a.real; + theta = 0; + } else { + r = -a.real; + theta = atan2f(0.0, -1.0); + } + } else { + r = __Pyx_c_abs_float(a); + theta = atan2f(a.imag, a.real); + } + lnr = logf(r); + z_r = expf(lnr * b.real - theta * b.imag); + z_theta = theta * b.real + lnr * b.imag; + z.real = z_r * cosf(z_theta); + z.imag = z_r * sinf(z_theta); + return z; + } + #endif +#endif + +/* Declarations */ +#if CYTHON_CCOMPLEX + #ifdef __cplusplus + static CYTHON_INLINE __pyx_t_double_complex __pyx_t_double_complex_from_parts(double x, double y) { + return ::std::complex< double >(x, y); + } + #else + static CYTHON_INLINE __pyx_t_double_complex __pyx_t_double_complex_from_parts(double x, double y) { + return x + y*(__pyx_t_double_complex)_Complex_I; + } + #endif +#else + static CYTHON_INLINE __pyx_t_double_complex __pyx_t_double_complex_from_parts(double x, double y) { + __pyx_t_double_complex z; + z.real = x; + z.imag = y; + return z; + } +#endif + +/* Arithmetic */ +#if CYTHON_CCOMPLEX +#else + static CYTHON_INLINE int __Pyx_c_eq_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + return (a.real == b.real) && (a.imag == b.imag); + } + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_sum_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + __pyx_t_double_complex z; + z.real = a.real + b.real; + z.imag = a.imag + b.imag; + return z; + } + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_diff_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + __pyx_t_double_complex z; + z.real = a.real - b.real; + z.imag = a.imag - b.imag; + return z; + } + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_prod_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + __pyx_t_double_complex z; + z.real = a.real * b.real - a.imag * b.imag; + z.imag = a.real * b.imag + a.imag * b.real; + return z; + } + #if 1 + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_quot_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + if (b.imag == 0) { + return __pyx_t_double_complex_from_parts(a.real / b.real, a.imag / b.real); + } else if (fabs(b.real) >= fabs(b.imag)) { + if (b.real == 0 && b.imag == 0) { + return __pyx_t_double_complex_from_parts(a.real / b.real, a.imag / b.imag); + } else { + double r = b.imag / b.real; + double s = (double)(1.0) / (b.real + b.imag * r); + return __pyx_t_double_complex_from_parts( + (a.real + a.imag * r) * s, (a.imag - a.real * r) * s); + } + } else { + double r = b.real / b.imag; + double s = (double)(1.0) / (b.imag + b.real * r); + return __pyx_t_double_complex_from_parts( + (a.real * r + a.imag) * s, (a.imag * r - a.real) * s); + } + } + #else + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_quot_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + if (b.imag == 0) { + return __pyx_t_double_complex_from_parts(a.real / b.real, a.imag / b.real); + } else { + double denom = b.real * b.real + b.imag * b.imag; + return __pyx_t_double_complex_from_parts( + (a.real * b.real + a.imag * b.imag) / denom, + (a.imag * b.real - a.real * b.imag) / denom); + } + } + #endif + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_neg_double(__pyx_t_double_complex a) { + __pyx_t_double_complex z; + z.real = -a.real; + z.imag = -a.imag; + return z; + } + static CYTHON_INLINE int __Pyx_c_is_zero_double(__pyx_t_double_complex a) { + return (a.real == 0) && (a.imag == 0); + } + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_conj_double(__pyx_t_double_complex a) { + __pyx_t_double_complex z; + z.real = a.real; + z.imag = -a.imag; + return z; + } + #if 1 + static CYTHON_INLINE double __Pyx_c_abs_double(__pyx_t_double_complex z) { + #if !defined(HAVE_HYPOT) || defined(_MSC_VER) + return sqrt(z.real*z.real + z.imag*z.imag); + #else + return hypot(z.real, z.imag); + #endif + } + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_pow_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + __pyx_t_double_complex z; + double r, lnr, theta, z_r, z_theta; + if (b.imag == 0 && b.real == (int)b.real) { + if (b.real < 0) { + double denom = a.real * a.real + a.imag * a.imag; + a.real = a.real / denom; + a.imag = -a.imag / denom; + b.real = -b.real; + } + switch ((int)b.real) { + case 0: + z.real = 1; + z.imag = 0; + return z; + case 1: + return a; + case 2: + return __Pyx_c_prod_double(a, a); + case 3: + z = __Pyx_c_prod_double(a, a); + return __Pyx_c_prod_double(z, a); + case 4: + z = __Pyx_c_prod_double(a, a); + return __Pyx_c_prod_double(z, z); + } + } + if (a.imag == 0) { + if (a.real == 0) { + return a; + } else if (b.imag == 0) { + z.real = pow(a.real, b.real); + z.imag = 0; + return z; + } else if (a.real > 0) { + r = a.real; + theta = 0; + } else { + r = -a.real; + theta = atan2(0.0, -1.0); + } + } else { + r = __Pyx_c_abs_double(a); + theta = atan2(a.imag, a.real); + } + lnr = log(r); + z_r = exp(lnr * b.real - theta * b.imag); + z_theta = theta * b.real + lnr * b.imag; + z.real = z_r * cos(z_theta); + z.imag = z_r * sin(z_theta); + return z; + } + #endif +#endif + +/* CIntToPy */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_enum__NPY_TYPES(enum NPY_TYPES value) { + const enum NPY_TYPES neg_one = (enum NPY_TYPES) ((enum NPY_TYPES) 0 - (enum NPY_TYPES) 1), const_zero = (enum NPY_TYPES) 0; + const int is_unsigned = neg_one > const_zero; + if (is_unsigned) { + if (sizeof(enum NPY_TYPES) < sizeof(long)) { + return PyInt_FromLong((long) value); + } else if (sizeof(enum NPY_TYPES) <= sizeof(unsigned long)) { + return PyLong_FromUnsignedLong((unsigned long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(enum NPY_TYPES) <= sizeof(unsigned PY_LONG_LONG)) { + return PyLong_FromUnsignedLongLong((unsigned PY_LONG_LONG) value); +#endif + } + } else { + if (sizeof(enum NPY_TYPES) <= sizeof(long)) { + return PyInt_FromLong((long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(enum NPY_TYPES) <= sizeof(PY_LONG_LONG)) { + return PyLong_FromLongLong((PY_LONG_LONG) value); +#endif + } + } + { + int one = 1; int little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&value; + return _PyLong_FromByteArray(bytes, sizeof(enum NPY_TYPES), + little, !is_unsigned); + } +} + +/* CIntFromPy */ +static CYTHON_INLINE int __Pyx_PyInt_As_int(PyObject *x) { + const int neg_one = (int) ((int) 0 - (int) 1), const_zero = (int) 0; + const int is_unsigned = neg_one > const_zero; +#if PY_MAJOR_VERSION < 3 + if (likely(PyInt_Check(x))) { + if (sizeof(int) < sizeof(long)) { + __PYX_VERIFY_RETURN_INT(int, long, PyInt_AS_LONG(x)) + } else { + long val = PyInt_AS_LONG(x); + if (is_unsigned && unlikely(val < 0)) { + goto raise_neg_overflow; + } + return (int) val; + } + } else +#endif + if (likely(PyLong_Check(x))) { + if (is_unsigned) { +#if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)x)->ob_digit; + switch (Py_SIZE(x)) { + case 0: return (int) 0; + case 1: __PYX_VERIFY_RETURN_INT(int, digit, digits[0]) + case 2: + if (8 * sizeof(int) > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, unsigned long, (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) >= 2 * PyLong_SHIFT) { + return (int) (((((int)digits[1]) << PyLong_SHIFT) | (int)digits[0])); + } + } + break; + case 3: + if (8 * sizeof(int) > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, unsigned long, (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) >= 3 * PyLong_SHIFT) { + return (int) (((((((int)digits[2]) << PyLong_SHIFT) | (int)digits[1]) << PyLong_SHIFT) | (int)digits[0])); + } + } + break; + case 4: + if (8 * sizeof(int) > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, unsigned long, (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) >= 4 * PyLong_SHIFT) { + return (int) (((((((((int)digits[3]) << PyLong_SHIFT) | (int)digits[2]) << PyLong_SHIFT) | (int)digits[1]) << PyLong_SHIFT) | (int)digits[0])); + } + } + break; + } +#endif +#if CYTHON_COMPILING_IN_CPYTHON + if (unlikely(Py_SIZE(x) < 0)) { + goto raise_neg_overflow; + } +#else + { + int result = PyObject_RichCompareBool(x, Py_False, Py_LT); + if (unlikely(result < 0)) + return (int) -1; + if (unlikely(result == 1)) + goto raise_neg_overflow; + } +#endif + if (sizeof(int) <= sizeof(unsigned long)) { + __PYX_VERIFY_RETURN_INT_EXC(int, unsigned long, PyLong_AsUnsignedLong(x)) +#ifdef HAVE_LONG_LONG + } else if (sizeof(int) <= sizeof(unsigned PY_LONG_LONG)) { + __PYX_VERIFY_RETURN_INT_EXC(int, unsigned PY_LONG_LONG, PyLong_AsUnsignedLongLong(x)) +#endif + } + } else { +#if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)x)->ob_digit; + switch (Py_SIZE(x)) { + case 0: return (int) 0; + case -1: __PYX_VERIFY_RETURN_INT(int, sdigit, (sdigit) (-(sdigit)digits[0])) + case 1: __PYX_VERIFY_RETURN_INT(int, digit, +digits[0]) + case -2: + if (8 * sizeof(int) - 1 > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, long, -(long) (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) - 1 > 2 * PyLong_SHIFT) { + return (int) (((int)-1)*(((((int)digits[1]) << PyLong_SHIFT) | (int)digits[0]))); + } + } + break; + case 2: + if (8 * sizeof(int) > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, unsigned long, (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) - 1 > 2 * PyLong_SHIFT) { + return (int) ((((((int)digits[1]) << PyLong_SHIFT) | (int)digits[0]))); + } + } + break; + case -3: + if (8 * sizeof(int) - 1 > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, long, -(long) (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) - 1 > 3 * PyLong_SHIFT) { + return (int) (((int)-1)*(((((((int)digits[2]) << PyLong_SHIFT) | (int)digits[1]) << PyLong_SHIFT) | (int)digits[0]))); + } + } + break; + case 3: + if (8 * sizeof(int) > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, unsigned long, (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) - 1 > 3 * PyLong_SHIFT) { + return (int) ((((((((int)digits[2]) << PyLong_SHIFT) | (int)digits[1]) << PyLong_SHIFT) | (int)digits[0]))); + } + } + break; + case -4: + if (8 * sizeof(int) - 1 > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, long, -(long) (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) - 1 > 4 * PyLong_SHIFT) { + return (int) (((int)-1)*(((((((((int)digits[3]) << PyLong_SHIFT) | (int)digits[2]) << PyLong_SHIFT) | (int)digits[1]) << PyLong_SHIFT) | (int)digits[0]))); + } + } + break; + case 4: + if (8 * sizeof(int) > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, unsigned long, (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) - 1 > 4 * PyLong_SHIFT) { + return (int) ((((((((((int)digits[3]) << PyLong_SHIFT) | (int)digits[2]) << PyLong_SHIFT) | (int)digits[1]) << PyLong_SHIFT) | (int)digits[0]))); + } + } + break; + } +#endif + if (sizeof(int) <= sizeof(long)) { + __PYX_VERIFY_RETURN_INT_EXC(int, long, PyLong_AsLong(x)) +#ifdef HAVE_LONG_LONG + } else if (sizeof(int) <= sizeof(PY_LONG_LONG)) { + __PYX_VERIFY_RETURN_INT_EXC(int, PY_LONG_LONG, PyLong_AsLongLong(x)) +#endif + } + } + { +#if CYTHON_COMPILING_IN_PYPY && !defined(_PyLong_AsByteArray) + PyErr_SetString(PyExc_RuntimeError, + "_PyLong_AsByteArray() not available in PyPy, cannot convert large numbers"); +#else + int val; + PyObject *v = __Pyx_PyNumber_IntOrLong(x); + #if PY_MAJOR_VERSION < 3 + if (likely(v) && !PyLong_Check(v)) { + PyObject *tmp = v; + v = PyNumber_Long(tmp); + Py_DECREF(tmp); + } + #endif + if (likely(v)) { + int one = 1; int is_little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&val; + int ret = _PyLong_AsByteArray((PyLongObject *)v, + bytes, sizeof(val), + is_little, !is_unsigned); + Py_DECREF(v); + if (likely(!ret)) + return val; + } +#endif + return (int) -1; + } + } else { + int val; + PyObject *tmp = __Pyx_PyNumber_IntOrLong(x); + if (!tmp) return (int) -1; + val = __Pyx_PyInt_As_int(tmp); + Py_DECREF(tmp); + return val; + } +raise_overflow: + PyErr_SetString(PyExc_OverflowError, + "value too large to convert to int"); + return (int) -1; +raise_neg_overflow: + PyErr_SetString(PyExc_OverflowError, + "can't convert negative value to int"); + return (int) -1; +} + +/* CIntFromPy */ +static CYTHON_INLINE npy_uint32 __Pyx_PyInt_As_npy_uint32(PyObject *x) { + const npy_uint32 neg_one = (npy_uint32) ((npy_uint32) 0 - (npy_uint32) 1), const_zero = (npy_uint32) 0; + const int is_unsigned = neg_one > const_zero; +#if PY_MAJOR_VERSION < 3 + if (likely(PyInt_Check(x))) { + if (sizeof(npy_uint32) < sizeof(long)) { + __PYX_VERIFY_RETURN_INT(npy_uint32, long, PyInt_AS_LONG(x)) + } else { + long val = PyInt_AS_LONG(x); + if (is_unsigned && unlikely(val < 0)) { + goto raise_neg_overflow; + } + return (npy_uint32) val; + } + } else +#endif + if (likely(PyLong_Check(x))) { + if (is_unsigned) { +#if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)x)->ob_digit; + switch (Py_SIZE(x)) { + case 0: return (npy_uint32) 0; + case 1: __PYX_VERIFY_RETURN_INT(npy_uint32, digit, digits[0]) + case 2: + if (8 * sizeof(npy_uint32) > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, unsigned long, (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) >= 2 * PyLong_SHIFT) { + return (npy_uint32) (((((npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0])); + } + } + break; + case 3: + if (8 * sizeof(npy_uint32) > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, unsigned long, (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) >= 3 * PyLong_SHIFT) { + return (npy_uint32) (((((((npy_uint32)digits[2]) << PyLong_SHIFT) | (npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0])); + } + } + break; + case 4: + if (8 * sizeof(npy_uint32) > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, unsigned long, (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) >= 4 * PyLong_SHIFT) { + return (npy_uint32) (((((((((npy_uint32)digits[3]) << PyLong_SHIFT) | (npy_uint32)digits[2]) << PyLong_SHIFT) | (npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0])); + } + } + break; + } +#endif +#if CYTHON_COMPILING_IN_CPYTHON + if (unlikely(Py_SIZE(x) < 0)) { + goto raise_neg_overflow; + } +#else + { + int result = PyObject_RichCompareBool(x, Py_False, Py_LT); + if (unlikely(result < 0)) + return (npy_uint32) -1; + if (unlikely(result == 1)) + goto raise_neg_overflow; + } +#endif + if (sizeof(npy_uint32) <= sizeof(unsigned long)) { + __PYX_VERIFY_RETURN_INT_EXC(npy_uint32, unsigned long, PyLong_AsUnsignedLong(x)) +#ifdef HAVE_LONG_LONG + } else if (sizeof(npy_uint32) <= sizeof(unsigned PY_LONG_LONG)) { + __PYX_VERIFY_RETURN_INT_EXC(npy_uint32, unsigned PY_LONG_LONG, PyLong_AsUnsignedLongLong(x)) +#endif + } + } else { +#if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)x)->ob_digit; + switch (Py_SIZE(x)) { + case 0: return (npy_uint32) 0; + case -1: __PYX_VERIFY_RETURN_INT(npy_uint32, sdigit, (sdigit) (-(sdigit)digits[0])) + case 1: __PYX_VERIFY_RETURN_INT(npy_uint32, digit, +digits[0]) + case -2: + if (8 * sizeof(npy_uint32) - 1 > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, long, -(long) (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) - 1 > 2 * PyLong_SHIFT) { + return (npy_uint32) (((npy_uint32)-1)*(((((npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0]))); + } + } + break; + case 2: + if (8 * sizeof(npy_uint32) > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, unsigned long, (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) - 1 > 2 * PyLong_SHIFT) { + return (npy_uint32) ((((((npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0]))); + } + } + break; + case -3: + if (8 * sizeof(npy_uint32) - 1 > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, long, -(long) (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) - 1 > 3 * PyLong_SHIFT) { + return (npy_uint32) (((npy_uint32)-1)*(((((((npy_uint32)digits[2]) << PyLong_SHIFT) | (npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0]))); + } + } + break; + case 3: + if (8 * sizeof(npy_uint32) > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, unsigned long, (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) - 1 > 3 * PyLong_SHIFT) { + return (npy_uint32) ((((((((npy_uint32)digits[2]) << PyLong_SHIFT) | (npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0]))); + } + } + break; + case -4: + if (8 * sizeof(npy_uint32) - 1 > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, long, -(long) (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) - 1 > 4 * PyLong_SHIFT) { + return (npy_uint32) (((npy_uint32)-1)*(((((((((npy_uint32)digits[3]) << PyLong_SHIFT) | (npy_uint32)digits[2]) << PyLong_SHIFT) | (npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0]))); + } + } + break; + case 4: + if (8 * sizeof(npy_uint32) > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, unsigned long, (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) - 1 > 4 * PyLong_SHIFT) { + return (npy_uint32) ((((((((((npy_uint32)digits[3]) << PyLong_SHIFT) | (npy_uint32)digits[2]) << PyLong_SHIFT) | (npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0]))); + } + } + break; + } +#endif + if (sizeof(npy_uint32) <= sizeof(long)) { + __PYX_VERIFY_RETURN_INT_EXC(npy_uint32, long, PyLong_AsLong(x)) +#ifdef HAVE_LONG_LONG + } else if (sizeof(npy_uint32) <= sizeof(PY_LONG_LONG)) { + __PYX_VERIFY_RETURN_INT_EXC(npy_uint32, PY_LONG_LONG, PyLong_AsLongLong(x)) +#endif + } + } + { +#if CYTHON_COMPILING_IN_PYPY && !defined(_PyLong_AsByteArray) + PyErr_SetString(PyExc_RuntimeError, + "_PyLong_AsByteArray() not available in PyPy, cannot convert large numbers"); +#else + npy_uint32 val; + PyObject *v = __Pyx_PyNumber_IntOrLong(x); + #if PY_MAJOR_VERSION < 3 + if (likely(v) && !PyLong_Check(v)) { + PyObject *tmp = v; + v = PyNumber_Long(tmp); + Py_DECREF(tmp); + } + #endif + if (likely(v)) { + int one = 1; int is_little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&val; + int ret = _PyLong_AsByteArray((PyLongObject *)v, + bytes, sizeof(val), + is_little, !is_unsigned); + Py_DECREF(v); + if (likely(!ret)) + return val; + } +#endif + return (npy_uint32) -1; + } + } else { + npy_uint32 val; + PyObject *tmp = __Pyx_PyNumber_IntOrLong(x); + if (!tmp) return (npy_uint32) -1; + val = __Pyx_PyInt_As_npy_uint32(tmp); + Py_DECREF(tmp); + return val; + } +raise_overflow: + PyErr_SetString(PyExc_OverflowError, + "value too large to convert to npy_uint32"); + return (npy_uint32) -1; +raise_neg_overflow: + PyErr_SetString(PyExc_OverflowError, + "can't convert negative value to npy_uint32"); + return (npy_uint32) -1; +} + +/* CIntFromPy */ +static CYTHON_INLINE long __Pyx_PyInt_As_long(PyObject *x) { + const long neg_one = (long) ((long) 0 - (long) 1), const_zero = (long) 0; + const int is_unsigned = neg_one > const_zero; +#if PY_MAJOR_VERSION < 3 + if (likely(PyInt_Check(x))) { + if (sizeof(long) < sizeof(long)) { + __PYX_VERIFY_RETURN_INT(long, long, PyInt_AS_LONG(x)) + } else { + long val = PyInt_AS_LONG(x); + if (is_unsigned && unlikely(val < 0)) { + goto raise_neg_overflow; + } + return (long) val; + } + } else +#endif + if (likely(PyLong_Check(x))) { + if (is_unsigned) { +#if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)x)->ob_digit; + switch (Py_SIZE(x)) { + case 0: return (long) 0; + case 1: __PYX_VERIFY_RETURN_INT(long, digit, digits[0]) + case 2: + if (8 * sizeof(long) > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, unsigned long, (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) >= 2 * PyLong_SHIFT) { + return (long) (((((long)digits[1]) << PyLong_SHIFT) | (long)digits[0])); + } + } + break; + case 3: + if (8 * sizeof(long) > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, unsigned long, (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) >= 3 * PyLong_SHIFT) { + return (long) (((((((long)digits[2]) << PyLong_SHIFT) | (long)digits[1]) << PyLong_SHIFT) | (long)digits[0])); + } + } + break; + case 4: + if (8 * sizeof(long) > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, unsigned long, (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) >= 4 * PyLong_SHIFT) { + return (long) (((((((((long)digits[3]) << PyLong_SHIFT) | (long)digits[2]) << PyLong_SHIFT) | (long)digits[1]) << PyLong_SHIFT) | (long)digits[0])); + } + } + break; + } +#endif +#if CYTHON_COMPILING_IN_CPYTHON + if (unlikely(Py_SIZE(x) < 0)) { + goto raise_neg_overflow; + } +#else + { + int result = PyObject_RichCompareBool(x, Py_False, Py_LT); + if (unlikely(result < 0)) + return (long) -1; + if (unlikely(result == 1)) + goto raise_neg_overflow; + } +#endif + if (sizeof(long) <= sizeof(unsigned long)) { + __PYX_VERIFY_RETURN_INT_EXC(long, unsigned long, PyLong_AsUnsignedLong(x)) +#ifdef HAVE_LONG_LONG + } else if (sizeof(long) <= sizeof(unsigned PY_LONG_LONG)) { + __PYX_VERIFY_RETURN_INT_EXC(long, unsigned PY_LONG_LONG, PyLong_AsUnsignedLongLong(x)) +#endif + } + } else { +#if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)x)->ob_digit; + switch (Py_SIZE(x)) { + case 0: return (long) 0; + case -1: __PYX_VERIFY_RETURN_INT(long, sdigit, (sdigit) (-(sdigit)digits[0])) + case 1: __PYX_VERIFY_RETURN_INT(long, digit, +digits[0]) + case -2: + if (8 * sizeof(long) - 1 > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, long, -(long) (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) - 1 > 2 * PyLong_SHIFT) { + return (long) (((long)-1)*(((((long)digits[1]) << PyLong_SHIFT) | (long)digits[0]))); + } + } + break; + case 2: + if (8 * sizeof(long) > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, unsigned long, (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) - 1 > 2 * PyLong_SHIFT) { + return (long) ((((((long)digits[1]) << PyLong_SHIFT) | (long)digits[0]))); + } + } + break; + case -3: + if (8 * sizeof(long) - 1 > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, long, -(long) (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) - 1 > 3 * PyLong_SHIFT) { + return (long) (((long)-1)*(((((((long)digits[2]) << PyLong_SHIFT) | (long)digits[1]) << PyLong_SHIFT) | (long)digits[0]))); + } + } + break; + case 3: + if (8 * sizeof(long) > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, unsigned long, (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) - 1 > 3 * PyLong_SHIFT) { + return (long) ((((((((long)digits[2]) << PyLong_SHIFT) | (long)digits[1]) << PyLong_SHIFT) | (long)digits[0]))); + } + } + break; + case -4: + if (8 * sizeof(long) - 1 > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, long, -(long) (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) - 1 > 4 * PyLong_SHIFT) { + return (long) (((long)-1)*(((((((((long)digits[3]) << PyLong_SHIFT) | (long)digits[2]) << PyLong_SHIFT) | (long)digits[1]) << PyLong_SHIFT) | (long)digits[0]))); + } + } + break; + case 4: + if (8 * sizeof(long) > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, unsigned long, (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) - 1 > 4 * PyLong_SHIFT) { + return (long) ((((((((((long)digits[3]) << PyLong_SHIFT) | (long)digits[2]) << PyLong_SHIFT) | (long)digits[1]) << PyLong_SHIFT) | (long)digits[0]))); + } + } + break; + } +#endif + if (sizeof(long) <= sizeof(long)) { + __PYX_VERIFY_RETURN_INT_EXC(long, long, PyLong_AsLong(x)) +#ifdef HAVE_LONG_LONG + } else if (sizeof(long) <= sizeof(PY_LONG_LONG)) { + __PYX_VERIFY_RETURN_INT_EXC(long, PY_LONG_LONG, PyLong_AsLongLong(x)) +#endif + } + } + { +#if CYTHON_COMPILING_IN_PYPY && !defined(_PyLong_AsByteArray) + PyErr_SetString(PyExc_RuntimeError, + "_PyLong_AsByteArray() not available in PyPy, cannot convert large numbers"); +#else + long val; + PyObject *v = __Pyx_PyNumber_IntOrLong(x); + #if PY_MAJOR_VERSION < 3 + if (likely(v) && !PyLong_Check(v)) { + PyObject *tmp = v; + v = PyNumber_Long(tmp); + Py_DECREF(tmp); + } + #endif + if (likely(v)) { + int one = 1; int is_little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&val; + int ret = _PyLong_AsByteArray((PyLongObject *)v, + bytes, sizeof(val), + is_little, !is_unsigned); + Py_DECREF(v); + if (likely(!ret)) + return val; + } +#endif + return (long) -1; + } + } else { + long val; + PyObject *tmp = __Pyx_PyNumber_IntOrLong(x); + if (!tmp) return (long) -1; + val = __Pyx_PyInt_As_long(tmp); + Py_DECREF(tmp); + return val; + } +raise_overflow: + PyErr_SetString(PyExc_OverflowError, + "value too large to convert to long"); + return (long) -1; +raise_neg_overflow: + PyErr_SetString(PyExc_OverflowError, + "can't convert negative value to long"); + return (long) -1; +} + +/* FastTypeChecks */ +#if CYTHON_COMPILING_IN_CPYTHON +static int __Pyx_InBases(PyTypeObject *a, PyTypeObject *b) { + while (a) { + a = a->tp_base; + if (a == b) + return 1; + } + return b == &PyBaseObject_Type; +} +static CYTHON_INLINE int __Pyx_IsSubtype(PyTypeObject *a, PyTypeObject *b) { + PyObject *mro; + if (a == b) return 1; + mro = a->tp_mro; + if (likely(mro)) { + Py_ssize_t i, n; + n = PyTuple_GET_SIZE(mro); + for (i = 0; i < n; i++) { + if (PyTuple_GET_ITEM(mro, i) == (PyObject *)b) + return 1; + } + return 0; + } + return __Pyx_InBases(a, b); +} +#if PY_MAJOR_VERSION == 2 +static int __Pyx_inner_PyErr_GivenExceptionMatches2(PyObject *err, PyObject* exc_type1, PyObject* exc_type2) { + PyObject *exception, *value, *tb; + int res; + __Pyx_PyThreadState_declare + __Pyx_PyThreadState_assign + __Pyx_ErrFetch(&exception, &value, &tb); + res = exc_type1 ? PyObject_IsSubclass(err, exc_type1) : 0; + if (unlikely(res == -1)) { + PyErr_WriteUnraisable(err); + res = 0; + } + if (!res) { + res = PyObject_IsSubclass(err, exc_type2); + if (unlikely(res == -1)) { + PyErr_WriteUnraisable(err); + res = 0; + } + } + __Pyx_ErrRestore(exception, value, tb); + return res; +} +#else +static CYTHON_INLINE int __Pyx_inner_PyErr_GivenExceptionMatches2(PyObject *err, PyObject* exc_type1, PyObject *exc_type2) { + int res = exc_type1 ? __Pyx_IsSubtype((PyTypeObject*)err, (PyTypeObject*)exc_type1) : 0; + if (!res) { + res = __Pyx_IsSubtype((PyTypeObject*)err, (PyTypeObject*)exc_type2); + } + return res; +} +#endif +static int __Pyx_PyErr_GivenExceptionMatchesTuple(PyObject *exc_type, PyObject *tuple) { + Py_ssize_t i, n; + assert(PyExceptionClass_Check(exc_type)); + n = PyTuple_GET_SIZE(tuple); +#if PY_MAJOR_VERSION >= 3 + for (i=0; itp_setattro)) + return tp->tp_setattro(obj, attr_name, value); +#if PY_MAJOR_VERSION < 3 + if (likely(tp->tp_setattr)) + return tp->tp_setattr(obj, PyString_AS_STRING(attr_name), value); +#endif + return PyObject_SetAttr(obj, attr_name, value); +} +#endif + +/* VoidPtrExport */ +static int __Pyx_ExportVoidPtr(PyObject *name, void *p, const char *sig) { + PyObject *d; + PyObject *cobj = 0; + d = PyDict_GetItem(__pyx_d, __pyx_n_s_pyx_capi); + Py_XINCREF(d); + if (!d) { + d = PyDict_New(); + if (!d) + goto bad; + if (__Pyx_PyObject_SetAttrStr(__pyx_m, __pyx_n_s_pyx_capi, d) < 0) + goto bad; + } +#if PY_VERSION_HEX >= 0x02070000 + cobj = PyCapsule_New(p, sig, 0); +#else + cobj = PyCObject_FromVoidPtrAndDesc(p, (void *)sig, 0); +#endif + if (!cobj) + goto bad; + if (PyDict_SetItem(d, name, cobj) < 0) + goto bad; + Py_DECREF(cobj); + Py_DECREF(d); + return 0; +bad: + Py_XDECREF(cobj); + Py_XDECREF(d); + return -1; +} + +/* FunctionExport */ +static int __Pyx_ExportFunction(const char *name, void (*f)(void), const char *sig) { + PyObject *d = 0; + PyObject *cobj = 0; + union { + void (*fp)(void); + void *p; + } tmp; + d = PyObject_GetAttrString(__pyx_m, (char *)"__pyx_capi__"); + if (!d) { + PyErr_Clear(); + d = PyDict_New(); + if (!d) + goto bad; + Py_INCREF(d); + if (PyModule_AddObject(__pyx_m, (char *)"__pyx_capi__", d) < 0) + goto bad; + } + tmp.fp = f; +#if PY_VERSION_HEX >= 0x02070000 + cobj = PyCapsule_New(tmp.p, sig, 0); +#else + cobj = PyCObject_FromVoidPtrAndDesc(tmp.p, (void *)sig, 0); +#endif + if (!cobj) + goto bad; + if (PyDict_SetItemString(d, name, cobj) < 0) + goto bad; + Py_DECREF(cobj); + Py_DECREF(d); + return 0; +bad: + Py_XDECREF(cobj); + Py_XDECREF(d); + return -1; +} + +/* InitStrings */ +static int __Pyx_InitStrings(__Pyx_StringTabEntry *t) { + while (t->p) { + #if PY_MAJOR_VERSION < 3 + if (t->is_unicode) { + *t->p = PyUnicode_DecodeUTF8(t->s, t->n - 1, NULL); + } else if (t->intern) { + *t->p = PyString_InternFromString(t->s); + } else { + *t->p = PyString_FromStringAndSize(t->s, t->n - 1); + } + #else + if (t->is_unicode | t->is_str) { + if (t->intern) { + *t->p = PyUnicode_InternFromString(t->s); + } else if (t->encoding) { + *t->p = PyUnicode_Decode(t->s, t->n - 1, t->encoding, NULL); + } else { + *t->p = PyUnicode_FromStringAndSize(t->s, t->n - 1); + } + } else { + *t->p = PyBytes_FromStringAndSize(t->s, t->n - 1); + } + #endif + if (!*t->p) + return -1; + if (PyObject_Hash(*t->p) == -1) + return -1; + ++t; + } + return 0; +} + +static CYTHON_INLINE PyObject* __Pyx_PyUnicode_FromString(const char* c_str) { + return __Pyx_PyUnicode_FromStringAndSize(c_str, (Py_ssize_t)strlen(c_str)); +} +static CYTHON_INLINE const char* __Pyx_PyObject_AsString(PyObject* o) { + Py_ssize_t ignore; + return __Pyx_PyObject_AsStringAndSize(o, &ignore); +} +#if __PYX_DEFAULT_STRING_ENCODING_IS_ASCII || __PYX_DEFAULT_STRING_ENCODING_IS_DEFAULT +#if !CYTHON_PEP393_ENABLED +static const char* __Pyx_PyUnicode_AsStringAndSize(PyObject* o, Py_ssize_t *length) { + char* defenc_c; + PyObject* defenc = _PyUnicode_AsDefaultEncodedString(o, NULL); + if (!defenc) return NULL; + defenc_c = PyBytes_AS_STRING(defenc); +#if __PYX_DEFAULT_STRING_ENCODING_IS_ASCII + { + char* end = defenc_c + PyBytes_GET_SIZE(defenc); + char* c; + for (c = defenc_c; c < end; c++) { + if ((unsigned char) (*c) >= 128) { + PyUnicode_AsASCIIString(o); + return NULL; + } + } + } +#endif + *length = PyBytes_GET_SIZE(defenc); + return defenc_c; +} +#else +static CYTHON_INLINE const char* __Pyx_PyUnicode_AsStringAndSize(PyObject* o, Py_ssize_t *length) { + if (unlikely(__Pyx_PyUnicode_READY(o) == -1)) return NULL; +#if __PYX_DEFAULT_STRING_ENCODING_IS_ASCII + if (likely(PyUnicode_IS_ASCII(o))) { + *length = PyUnicode_GET_LENGTH(o); + return PyUnicode_AsUTF8(o); + } else { + PyUnicode_AsASCIIString(o); + return NULL; + } +#else + return PyUnicode_AsUTF8AndSize(o, length); +#endif +} +#endif +#endif +static CYTHON_INLINE const char* __Pyx_PyObject_AsStringAndSize(PyObject* o, Py_ssize_t *length) { +#if __PYX_DEFAULT_STRING_ENCODING_IS_ASCII || __PYX_DEFAULT_STRING_ENCODING_IS_DEFAULT + if ( +#if PY_MAJOR_VERSION < 3 && __PYX_DEFAULT_STRING_ENCODING_IS_ASCII + __Pyx_sys_getdefaultencoding_not_ascii && +#endif + PyUnicode_Check(o)) { + return __Pyx_PyUnicode_AsStringAndSize(o, length); + } else +#endif +#if (!CYTHON_COMPILING_IN_PYPY) || (defined(PyByteArray_AS_STRING) && defined(PyByteArray_GET_SIZE)) + if (PyByteArray_Check(o)) { + *length = PyByteArray_GET_SIZE(o); + return PyByteArray_AS_STRING(o); + } else +#endif + { + char* result; + int r = PyBytes_AsStringAndSize(o, &result, length); + if (unlikely(r < 0)) { + return NULL; + } else { + return result; + } + } +} +static CYTHON_INLINE int __Pyx_PyObject_IsTrue(PyObject* x) { + int is_true = x == Py_True; + if (is_true | (x == Py_False) | (x == Py_None)) return is_true; + else return PyObject_IsTrue(x); +} +static CYTHON_INLINE int __Pyx_PyObject_IsTrueAndDecref(PyObject* x) { + int retval; + if (unlikely(!x)) return -1; + retval = __Pyx_PyObject_IsTrue(x); + Py_DECREF(x); + return retval; +} +static PyObject* __Pyx_PyNumber_IntOrLongWrongResultType(PyObject* result, const char* type_name) { +#if PY_MAJOR_VERSION >= 3 + if (PyLong_Check(result)) { + if (PyErr_WarnFormat(PyExc_DeprecationWarning, 1, + "__int__ returned non-int (type %.200s). " + "The ability to return an instance of a strict subclass of int " + "is deprecated, and may be removed in a future version of Python.", + Py_TYPE(result)->tp_name)) { + Py_DECREF(result); + return NULL; + } + return result; + } +#endif + PyErr_Format(PyExc_TypeError, + "__%.4s__ returned non-%.4s (type %.200s)", + type_name, type_name, Py_TYPE(result)->tp_name); + Py_DECREF(result); + return NULL; +} +static CYTHON_INLINE PyObject* __Pyx_PyNumber_IntOrLong(PyObject* x) { +#if CYTHON_USE_TYPE_SLOTS + PyNumberMethods *m; +#endif + const char *name = NULL; + PyObject *res = NULL; +#if PY_MAJOR_VERSION < 3 + if (likely(PyInt_Check(x) || PyLong_Check(x))) +#else + if (likely(PyLong_Check(x))) +#endif + return __Pyx_NewRef(x); +#if CYTHON_USE_TYPE_SLOTS + m = Py_TYPE(x)->tp_as_number; + #if PY_MAJOR_VERSION < 3 + if (m && m->nb_int) { + name = "int"; + res = m->nb_int(x); + } + else if (m && m->nb_long) { + name = "long"; + res = m->nb_long(x); + } + #else + if (likely(m && m->nb_int)) { + name = "int"; + res = m->nb_int(x); + } + #endif +#else + if (!PyBytes_CheckExact(x) && !PyUnicode_CheckExact(x)) { + res = PyNumber_Int(x); + } +#endif + if (likely(res)) { +#if PY_MAJOR_VERSION < 3 + if (unlikely(!PyInt_Check(res) && !PyLong_Check(res))) { +#else + if (unlikely(!PyLong_CheckExact(res))) { +#endif + return __Pyx_PyNumber_IntOrLongWrongResultType(res, name); + } + } + else if (!PyErr_Occurred()) { + PyErr_SetString(PyExc_TypeError, + "an integer is required"); + } + return res; +} +static CYTHON_INLINE Py_ssize_t __Pyx_PyIndex_AsSsize_t(PyObject* b) { + Py_ssize_t ival; + PyObject *x; +#if PY_MAJOR_VERSION < 3 + if (likely(PyInt_CheckExact(b))) { + if (sizeof(Py_ssize_t) >= sizeof(long)) + return PyInt_AS_LONG(b); + else + return PyInt_AsSsize_t(b); + } +#endif + if (likely(PyLong_CheckExact(b))) { + #if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)b)->ob_digit; + const Py_ssize_t size = Py_SIZE(b); + if (likely(__Pyx_sst_abs(size) <= 1)) { + ival = likely(size) ? digits[0] : 0; + if (size == -1) ival = -ival; + return ival; + } else { + switch (size) { + case 2: + if (8 * sizeof(Py_ssize_t) > 2 * PyLong_SHIFT) { + return (Py_ssize_t) (((((size_t)digits[1]) << PyLong_SHIFT) | (size_t)digits[0])); + } + break; + case -2: + if (8 * sizeof(Py_ssize_t) > 2 * PyLong_SHIFT) { + return -(Py_ssize_t) (((((size_t)digits[1]) << PyLong_SHIFT) | (size_t)digits[0])); + } + break; + case 3: + if (8 * sizeof(Py_ssize_t) > 3 * PyLong_SHIFT) { + return (Py_ssize_t) (((((((size_t)digits[2]) << PyLong_SHIFT) | (size_t)digits[1]) << PyLong_SHIFT) | (size_t)digits[0])); + } + break; + case -3: + if (8 * sizeof(Py_ssize_t) > 3 * PyLong_SHIFT) { + return -(Py_ssize_t) (((((((size_t)digits[2]) << PyLong_SHIFT) | (size_t)digits[1]) << PyLong_SHIFT) | (size_t)digits[0])); + } + break; + case 4: + if (8 * sizeof(Py_ssize_t) > 4 * PyLong_SHIFT) { + return (Py_ssize_t) (((((((((size_t)digits[3]) << PyLong_SHIFT) | (size_t)digits[2]) << PyLong_SHIFT) | (size_t)digits[1]) << PyLong_SHIFT) | (size_t)digits[0])); + } + break; + case -4: + if (8 * sizeof(Py_ssize_t) > 4 * PyLong_SHIFT) { + return -(Py_ssize_t) (((((((((size_t)digits[3]) << PyLong_SHIFT) | (size_t)digits[2]) << PyLong_SHIFT) | (size_t)digits[1]) << PyLong_SHIFT) | (size_t)digits[0])); + } + break; + } + } + #endif + return PyLong_AsSsize_t(b); + } + x = PyNumber_Index(b); + if (!x) return -1; + ival = PyInt_AsSsize_t(x); + Py_DECREF(x); + return ival; +} +static CYTHON_INLINE PyObject * __Pyx_PyBool_FromLong(long b) { + return b ? __Pyx_NewRef(Py_True) : __Pyx_NewRef(Py_False); +} +static CYTHON_INLINE PyObject * __Pyx_PyInt_FromSize_t(size_t ival) { + return PyInt_FromSize_t(ival); +} + + +#endif /* Py_PYTHON_H */ diff --git a/fse/models/average_inner.pxd b/fse/models/average_inner.pxd index c725ea4..1cdc509 100644 --- a/fse/models/average_inner.pxd +++ b/fse/models/average_inner.pxd @@ -22,6 +22,12 @@ ctypedef void (*sscal_ptr) (const int *N, const float *alpha, const float *X, co cdef saxpy_ptr saxpy cdef sscal_ptr sscal +cdef int ONE = 1 +cdef int ZERO = 0 + +cdef REAL_t ONEF = 1.0 +cdef REAL_t ZEROF = 0.0 + DEF MAX_WORDS = 10000 DEF MAX_NGRAMS = 40 @@ -30,6 +36,7 @@ cdef struct BaseSentenceVecsConfig: # Vectors REAL_t *mem + REAL_t *mem2 REAL_t *word_vectors REAL_t *word_weights REAL_t *sentence_vectors @@ -45,6 +52,7 @@ cdef struct FTSentenceVecsConfig: # Vectors REAL_t *mem + REAL_t *mem2 REAL_t *word_vectors # Note: these will be the vocab vectors, not wv.vectors REAL_t *ngram_vectors REAL_t *word_weights @@ -61,5 +69,7 @@ cdef struct FTSentenceVecsConfig: uINT_t *subwords_idx cdef init_base_s2v_config(BaseSentenceVecsConfig *c, model, target, memory) +cdef init_ft_s2v_config(FTSentenceVecsConfig *c, model, target, memory) -cdef init_ft_s2v_config(FTSentenceVecsConfig *c, model, target, memory) \ No newline at end of file +cdef object populate_base_s2v_config(BaseSentenceVecsConfig *c, vocab, indexed_sentences) +cdef object populate_ft_s2v_config(FTSentenceVecsConfig *c, vocab, indexed_sentences) \ No newline at end of file diff --git a/fse/models/average_inner.pyx b/fse/models/average_inner.pyx index 28589ad..5694e2f 100644 --- a/fse/models/average_inner.pyx +++ b/fse/models/average_inner.pyx @@ -18,7 +18,6 @@ cimport numpy as np from gensim.models._utils_any2vec import compute_ngrams_bytes, ft_hash_bytes from libc.string cimport memset -from libc.stdio cimport printf import scipy.linalg.blas as fblas @@ -47,13 +46,14 @@ cdef init_base_s2v_config(BaseSentenceVecsConfig *c, model, target, memory): The target array to write the averages to. memory : np.ndarray Private working memory for each worker. - Consists of 2 nd.arrays. + Consists of 3 nd.arrays. """ c[0].workers = model.workers c[0].size = model.sv.vector_size c[0].mem = (np.PyArray_DATA(memory[0])) + c[0].mem2 = (np.PyArray_DATA(memory[2])) c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors)) c[0].word_weights = (np.PyArray_DATA(model.word_weights)) @@ -73,7 +73,7 @@ cdef init_ft_s2v_config(FTSentenceVecsConfig *c, model, target, memory): The target array to write the averages to. memory : np.ndarray Private working memory for each worker. - Consists of 2 nd.arrays. + Consists of 3 nd.arrays. """ @@ -86,6 +86,7 @@ cdef init_ft_s2v_config(FTSentenceVecsConfig *c, model, target, memory): c[0].oov_weight = np.max(model.word_weights) c[0].mem = (np.PyArray_DATA(memory[0])) + c[0].mem2 = (np.PyArray_DATA(memory[2])) memory[1].fill(ZERO) # Reset the ngram storage before filling the struct c[0].subwords_idx = (np.PyArray_DATA(memory[1])) diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index 3dc4b42..673255d 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -534,8 +534,13 @@ def _get_thread_working_mem(self) -> tuple: """ mem = zeros_aligned(self.sv.vector_size, dtype=REAL) - oov_mem = zeros_aligned((self.batch_words, self.batch_ngrams), dtype=uINT) - return (mem, oov_mem) + if self.is_ft: + oov_mem = zeros_aligned((self.batch_words, self.batch_ngrams), dtype=uINT) + else: + # Save memory if not using FT + oov_mem = 0 + mem2 = zeros_aligned(self.sv.vector_size, dtype=REAL) + return (mem, oov_mem, mem2) def _do_train_job( self, data_iterable: List[tuple], target: ndarray, memory: tuple diff --git a/fse/models/pooling.py b/fse/models/pooling.py index a14f4e4..855f452 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -62,7 +62,7 @@ def train_pooling_np( ) -> [int, int]: """Training on a sequence of sentences and update the target ndarray. - Called internally from :meth:`~fse.models.average.Average._do_train_job`. + Called internally from :meth:`~fse.models.pooling.MaxPooling._do_train_job`. Warnings -------- @@ -220,21 +220,19 @@ def get_ft_vector(word: str) -> ndarray: return eff_sentences, eff_words -# try: -# from fse.models.average_inner import train_average_cy -# from fse.models.average_inner import ( -# FAST_VERSION, -# MAX_WORDS_IN_BATCH, -# MAX_NGRAMS_IN_BATCH, -# ) - -# train_average = train_average_cy -# except ImportError - -FAST_VERSION = -1 -MAX_WORDS_IN_BATCH = 10000 -MAX_NGRAMS_IN_BATCH = 40 -train_pooling = train_pooling_np +try: + from fse.models.pooling_inner import train_pooling_cy + from fse.models.pooling_inner import ( + FAST_VERSION, + MAX_WORDS_IN_BATCH, + MAX_NGRAMS_IN_BATCH, + ) + train_pooling = train_pooling_cy +except ImportError: + FAST_VERSION = -1 + MAX_WORDS_IN_BATCH = 10000 + MAX_NGRAMS_IN_BATCH = 40 + train_pooling = train_pooling_np class MaxPooling(BaseSentence2VecModel): diff --git a/fse/models/pooling_inner.c b/fse/models/pooling_inner.c new file mode 100644 index 0000000..d8a776c --- /dev/null +++ b/fse/models/pooling_inner.c @@ -0,0 +1,8350 @@ +/* Generated by Cython 0.29.14 */ + +/* BEGIN: Cython Metadata +{ + "distutils": { + "depends": [ + "/Users/oliverborchers/Library/Mobile Documents/com~apple~CloudDocs/Diss/Medium/Fast_Sentence_Embeddings/fse/models/voidptr.h" + ], + "extra_compile_args": [ + "-O2", + "-march=native" + ], + "include_dirs": [ + "/Users/oliverborchers/Library/Mobile Documents/com~apple~CloudDocs/Diss/Medium/Fast_Sentence_Embeddings/fse/models" + ], + "name": "pooling_inner", + "sources": [ + "pooling_inner.pyx" + ] + }, + "module_name": "pooling_inner" +} +END: Cython Metadata */ + +#define PY_SSIZE_T_CLEAN +#include "Python.h" +#ifndef Py_PYTHON_H + #error Python headers needed to compile C extensions, please install development version of Python. +#elif PY_VERSION_HEX < 0x02060000 || (0x03000000 <= PY_VERSION_HEX && PY_VERSION_HEX < 0x03030000) + #error Cython requires Python 2.6+ or Python 3.3+. +#else +#define CYTHON_ABI "0_29_14" +#define CYTHON_HEX_VERSION 0x001D0EF0 +#define CYTHON_FUTURE_DIVISION 0 +#include +#ifndef offsetof + #define offsetof(type, member) ( (size_t) & ((type*)0) -> member ) +#endif +#if !defined(WIN32) && !defined(MS_WINDOWS) + #ifndef __stdcall + #define __stdcall + #endif + #ifndef __cdecl + #define __cdecl + #endif + #ifndef __fastcall + #define __fastcall + #endif +#endif +#ifndef DL_IMPORT + #define DL_IMPORT(t) t +#endif +#ifndef DL_EXPORT + #define DL_EXPORT(t) t +#endif +#define __PYX_COMMA , +#ifndef HAVE_LONG_LONG + #if PY_VERSION_HEX >= 0x02070000 + #define HAVE_LONG_LONG + #endif +#endif +#ifndef PY_LONG_LONG + #define PY_LONG_LONG LONG_LONG +#endif +#ifndef Py_HUGE_VAL + #define Py_HUGE_VAL HUGE_VAL +#endif +#ifdef PYPY_VERSION + #define CYTHON_COMPILING_IN_PYPY 1 + #define CYTHON_COMPILING_IN_PYSTON 0 + #define CYTHON_COMPILING_IN_CPYTHON 0 + #undef CYTHON_USE_TYPE_SLOTS + #define CYTHON_USE_TYPE_SLOTS 0 + #undef CYTHON_USE_PYTYPE_LOOKUP + #define CYTHON_USE_PYTYPE_LOOKUP 0 + #if PY_VERSION_HEX < 0x03050000 + #undef CYTHON_USE_ASYNC_SLOTS + #define CYTHON_USE_ASYNC_SLOTS 0 + #elif !defined(CYTHON_USE_ASYNC_SLOTS) + #define CYTHON_USE_ASYNC_SLOTS 1 + #endif + #undef CYTHON_USE_PYLIST_INTERNALS + #define CYTHON_USE_PYLIST_INTERNALS 0 + #undef CYTHON_USE_UNICODE_INTERNALS + #define CYTHON_USE_UNICODE_INTERNALS 0 + #undef CYTHON_USE_UNICODE_WRITER + #define CYTHON_USE_UNICODE_WRITER 0 + #undef CYTHON_USE_PYLONG_INTERNALS + #define CYTHON_USE_PYLONG_INTERNALS 0 + #undef CYTHON_AVOID_BORROWED_REFS + #define CYTHON_AVOID_BORROWED_REFS 1 + #undef CYTHON_ASSUME_SAFE_MACROS + #define CYTHON_ASSUME_SAFE_MACROS 0 + #undef CYTHON_UNPACK_METHODS + #define CYTHON_UNPACK_METHODS 0 + #undef CYTHON_FAST_THREAD_STATE + #define CYTHON_FAST_THREAD_STATE 0 + #undef CYTHON_FAST_PYCALL + #define CYTHON_FAST_PYCALL 0 + #undef CYTHON_PEP489_MULTI_PHASE_INIT + #define CYTHON_PEP489_MULTI_PHASE_INIT 0 + #undef CYTHON_USE_TP_FINALIZE + #define CYTHON_USE_TP_FINALIZE 0 + #undef CYTHON_USE_DICT_VERSIONS + #define CYTHON_USE_DICT_VERSIONS 0 + #undef CYTHON_USE_EXC_INFO_STACK + #define CYTHON_USE_EXC_INFO_STACK 0 +#elif defined(PYSTON_VERSION) + #define CYTHON_COMPILING_IN_PYPY 0 + #define CYTHON_COMPILING_IN_PYSTON 1 + #define CYTHON_COMPILING_IN_CPYTHON 0 + #ifndef CYTHON_USE_TYPE_SLOTS + #define CYTHON_USE_TYPE_SLOTS 1 + #endif + #undef CYTHON_USE_PYTYPE_LOOKUP + #define CYTHON_USE_PYTYPE_LOOKUP 0 + #undef CYTHON_USE_ASYNC_SLOTS + #define CYTHON_USE_ASYNC_SLOTS 0 + #undef CYTHON_USE_PYLIST_INTERNALS + #define CYTHON_USE_PYLIST_INTERNALS 0 + #ifndef CYTHON_USE_UNICODE_INTERNALS + #define CYTHON_USE_UNICODE_INTERNALS 1 + #endif + #undef CYTHON_USE_UNICODE_WRITER + #define CYTHON_USE_UNICODE_WRITER 0 + #undef CYTHON_USE_PYLONG_INTERNALS + #define CYTHON_USE_PYLONG_INTERNALS 0 + #ifndef CYTHON_AVOID_BORROWED_REFS + #define CYTHON_AVOID_BORROWED_REFS 0 + #endif + #ifndef CYTHON_ASSUME_SAFE_MACROS + #define CYTHON_ASSUME_SAFE_MACROS 1 + #endif + #ifndef CYTHON_UNPACK_METHODS + #define CYTHON_UNPACK_METHODS 1 + #endif + #undef CYTHON_FAST_THREAD_STATE + #define CYTHON_FAST_THREAD_STATE 0 + #undef CYTHON_FAST_PYCALL + #define CYTHON_FAST_PYCALL 0 + #undef CYTHON_PEP489_MULTI_PHASE_INIT + #define CYTHON_PEP489_MULTI_PHASE_INIT 0 + #undef CYTHON_USE_TP_FINALIZE + #define CYTHON_USE_TP_FINALIZE 0 + #undef CYTHON_USE_DICT_VERSIONS + #define CYTHON_USE_DICT_VERSIONS 0 + #undef CYTHON_USE_EXC_INFO_STACK + #define CYTHON_USE_EXC_INFO_STACK 0 +#else + #define CYTHON_COMPILING_IN_PYPY 0 + #define CYTHON_COMPILING_IN_PYSTON 0 + #define CYTHON_COMPILING_IN_CPYTHON 1 + #ifndef CYTHON_USE_TYPE_SLOTS + #define CYTHON_USE_TYPE_SLOTS 1 + #endif + #if PY_VERSION_HEX < 0x02070000 + #undef CYTHON_USE_PYTYPE_LOOKUP + #define CYTHON_USE_PYTYPE_LOOKUP 0 + #elif !defined(CYTHON_USE_PYTYPE_LOOKUP) + #define CYTHON_USE_PYTYPE_LOOKUP 1 + #endif + #if PY_MAJOR_VERSION < 3 + #undef CYTHON_USE_ASYNC_SLOTS + #define CYTHON_USE_ASYNC_SLOTS 0 + #elif !defined(CYTHON_USE_ASYNC_SLOTS) + #define CYTHON_USE_ASYNC_SLOTS 1 + #endif + #if PY_VERSION_HEX < 0x02070000 + #undef CYTHON_USE_PYLONG_INTERNALS + #define CYTHON_USE_PYLONG_INTERNALS 0 + #elif !defined(CYTHON_USE_PYLONG_INTERNALS) + #define CYTHON_USE_PYLONG_INTERNALS 1 + #endif + #ifndef CYTHON_USE_PYLIST_INTERNALS + #define CYTHON_USE_PYLIST_INTERNALS 1 + #endif + #ifndef CYTHON_USE_UNICODE_INTERNALS + #define CYTHON_USE_UNICODE_INTERNALS 1 + #endif + #if PY_VERSION_HEX < 0x030300F0 + #undef CYTHON_USE_UNICODE_WRITER + #define CYTHON_USE_UNICODE_WRITER 0 + #elif !defined(CYTHON_USE_UNICODE_WRITER) + #define CYTHON_USE_UNICODE_WRITER 1 + #endif + #ifndef CYTHON_AVOID_BORROWED_REFS + #define CYTHON_AVOID_BORROWED_REFS 0 + #endif + #ifndef CYTHON_ASSUME_SAFE_MACROS + #define CYTHON_ASSUME_SAFE_MACROS 1 + #endif + #ifndef CYTHON_UNPACK_METHODS + #define CYTHON_UNPACK_METHODS 1 + #endif + #ifndef CYTHON_FAST_THREAD_STATE + #define CYTHON_FAST_THREAD_STATE 1 + #endif + #ifndef CYTHON_FAST_PYCALL + #define CYTHON_FAST_PYCALL 1 + #endif + #ifndef CYTHON_PEP489_MULTI_PHASE_INIT + #define CYTHON_PEP489_MULTI_PHASE_INIT (PY_VERSION_HEX >= 0x03050000) + #endif + #ifndef CYTHON_USE_TP_FINALIZE + #define CYTHON_USE_TP_FINALIZE (PY_VERSION_HEX >= 0x030400a1) + #endif + #ifndef CYTHON_USE_DICT_VERSIONS + #define CYTHON_USE_DICT_VERSIONS (PY_VERSION_HEX >= 0x030600B1) + #endif + #ifndef CYTHON_USE_EXC_INFO_STACK + #define CYTHON_USE_EXC_INFO_STACK (PY_VERSION_HEX >= 0x030700A3) + #endif +#endif +#if !defined(CYTHON_FAST_PYCCALL) +#define CYTHON_FAST_PYCCALL (CYTHON_FAST_PYCALL && PY_VERSION_HEX >= 0x030600B1) +#endif +#if CYTHON_USE_PYLONG_INTERNALS + #include "longintrepr.h" + #undef SHIFT + #undef BASE + #undef MASK + #ifdef SIZEOF_VOID_P + enum { __pyx_check_sizeof_voidp = 1 / (int)(SIZEOF_VOID_P == sizeof(void*)) }; + #endif +#endif +#ifndef __has_attribute + #define __has_attribute(x) 0 +#endif +#ifndef __has_cpp_attribute + #define __has_cpp_attribute(x) 0 +#endif +#ifndef CYTHON_RESTRICT + #if defined(__GNUC__) + #define CYTHON_RESTRICT __restrict__ + #elif defined(_MSC_VER) && _MSC_VER >= 1400 + #define CYTHON_RESTRICT __restrict + #elif defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L + #define CYTHON_RESTRICT restrict + #else + #define CYTHON_RESTRICT + #endif +#endif +#ifndef CYTHON_UNUSED +# if defined(__GNUC__) +# if !(defined(__cplusplus)) || (__GNUC__ > 3 || (__GNUC__ == 3 && __GNUC_MINOR__ >= 4)) +# define CYTHON_UNUSED __attribute__ ((__unused__)) +# else +# define CYTHON_UNUSED +# endif +# elif defined(__ICC) || (defined(__INTEL_COMPILER) && !defined(_MSC_VER)) +# define CYTHON_UNUSED __attribute__ ((__unused__)) +# else +# define CYTHON_UNUSED +# endif +#endif +#ifndef CYTHON_MAYBE_UNUSED_VAR +# if defined(__cplusplus) + template void CYTHON_MAYBE_UNUSED_VAR( const T& ) { } +# else +# define CYTHON_MAYBE_UNUSED_VAR(x) (void)(x) +# endif +#endif +#ifndef CYTHON_NCP_UNUSED +# if CYTHON_COMPILING_IN_CPYTHON +# define CYTHON_NCP_UNUSED +# else +# define CYTHON_NCP_UNUSED CYTHON_UNUSED +# endif +#endif +#define __Pyx_void_to_None(void_result) ((void)(void_result), Py_INCREF(Py_None), Py_None) +#ifdef _MSC_VER + #ifndef _MSC_STDINT_H_ + #if _MSC_VER < 1300 + typedef unsigned char uint8_t; + typedef unsigned int uint32_t; + #else + typedef unsigned __int8 uint8_t; + typedef unsigned __int32 uint32_t; + #endif + #endif +#else + #include +#endif +#ifndef CYTHON_FALLTHROUGH + #if defined(__cplusplus) && __cplusplus >= 201103L + #if __has_cpp_attribute(fallthrough) + #define CYTHON_FALLTHROUGH [[fallthrough]] + #elif __has_cpp_attribute(clang::fallthrough) + #define CYTHON_FALLTHROUGH [[clang::fallthrough]] + #elif __has_cpp_attribute(gnu::fallthrough) + #define CYTHON_FALLTHROUGH [[gnu::fallthrough]] + #endif + #endif + #ifndef CYTHON_FALLTHROUGH + #if __has_attribute(fallthrough) + #define CYTHON_FALLTHROUGH __attribute__((fallthrough)) + #else + #define CYTHON_FALLTHROUGH + #endif + #endif + #if defined(__clang__ ) && defined(__apple_build_version__) + #if __apple_build_version__ < 7000000 + #undef CYTHON_FALLTHROUGH + #define CYTHON_FALLTHROUGH + #endif + #endif +#endif + +#ifndef CYTHON_INLINE + #if defined(__clang__) + #define CYTHON_INLINE __inline__ __attribute__ ((__unused__)) + #elif defined(__GNUC__) + #define CYTHON_INLINE __inline__ + #elif defined(_MSC_VER) + #define CYTHON_INLINE __inline + #elif defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L + #define CYTHON_INLINE inline + #else + #define CYTHON_INLINE + #endif +#endif + +#if CYTHON_COMPILING_IN_PYPY && PY_VERSION_HEX < 0x02070600 && !defined(Py_OptimizeFlag) + #define Py_OptimizeFlag 0 +#endif +#define __PYX_BUILD_PY_SSIZE_T "n" +#define CYTHON_FORMAT_SSIZE_T "z" +#if PY_MAJOR_VERSION < 3 + #define __Pyx_BUILTIN_MODULE_NAME "__builtin__" + #define __Pyx_PyCode_New(a, k, l, s, f, code, c, n, v, fv, cell, fn, name, fline, lnos)\ + PyCode_New(a+k, l, s, f, code, c, n, v, fv, cell, fn, name, fline, lnos) + #define __Pyx_DefaultClassType PyClass_Type +#else + #define __Pyx_BUILTIN_MODULE_NAME "builtins" +#if PY_VERSION_HEX >= 0x030800A4 && PY_VERSION_HEX < 0x030800B2 + #define __Pyx_PyCode_New(a, k, l, s, f, code, c, n, v, fv, cell, fn, name, fline, lnos)\ + PyCode_New(a, 0, k, l, s, f, code, c, n, v, fv, cell, fn, name, fline, lnos) +#else + #define __Pyx_PyCode_New(a, k, l, s, f, code, c, n, v, fv, cell, fn, name, fline, lnos)\ + PyCode_New(a, k, l, s, f, code, c, n, v, fv, cell, fn, name, fline, lnos) +#endif + #define __Pyx_DefaultClassType PyType_Type +#endif +#ifndef Py_TPFLAGS_CHECKTYPES + #define Py_TPFLAGS_CHECKTYPES 0 +#endif +#ifndef Py_TPFLAGS_HAVE_INDEX + #define Py_TPFLAGS_HAVE_INDEX 0 +#endif +#ifndef Py_TPFLAGS_HAVE_NEWBUFFER + #define Py_TPFLAGS_HAVE_NEWBUFFER 0 +#endif +#ifndef Py_TPFLAGS_HAVE_FINALIZE + #define Py_TPFLAGS_HAVE_FINALIZE 0 +#endif +#ifndef METH_STACKLESS + #define METH_STACKLESS 0 +#endif +#if PY_VERSION_HEX <= 0x030700A3 || !defined(METH_FASTCALL) + #ifndef METH_FASTCALL + #define METH_FASTCALL 0x80 + #endif + typedef PyObject *(*__Pyx_PyCFunctionFast) (PyObject *self, PyObject *const *args, Py_ssize_t nargs); + typedef PyObject *(*__Pyx_PyCFunctionFastWithKeywords) (PyObject *self, PyObject *const *args, + Py_ssize_t nargs, PyObject *kwnames); +#else + #define __Pyx_PyCFunctionFast _PyCFunctionFast + #define __Pyx_PyCFunctionFastWithKeywords _PyCFunctionFastWithKeywords +#endif +#if CYTHON_FAST_PYCCALL +#define __Pyx_PyFastCFunction_Check(func)\ + ((PyCFunction_Check(func) && (METH_FASTCALL == (PyCFunction_GET_FLAGS(func) & ~(METH_CLASS | METH_STATIC | METH_COEXIST | METH_KEYWORDS | METH_STACKLESS))))) +#else +#define __Pyx_PyFastCFunction_Check(func) 0 +#endif +#if CYTHON_COMPILING_IN_PYPY && !defined(PyObject_Malloc) + #define PyObject_Malloc(s) PyMem_Malloc(s) + #define PyObject_Free(p) PyMem_Free(p) + #define PyObject_Realloc(p) PyMem_Realloc(p) +#endif +#if CYTHON_COMPILING_IN_CPYTHON && PY_VERSION_HEX < 0x030400A1 + #define PyMem_RawMalloc(n) PyMem_Malloc(n) + #define PyMem_RawRealloc(p, n) PyMem_Realloc(p, n) + #define PyMem_RawFree(p) PyMem_Free(p) +#endif +#if CYTHON_COMPILING_IN_PYSTON + #define __Pyx_PyCode_HasFreeVars(co) PyCode_HasFreeVars(co) + #define __Pyx_PyFrame_SetLineNumber(frame, lineno) PyFrame_SetLineNumber(frame, lineno) +#else + #define __Pyx_PyCode_HasFreeVars(co) (PyCode_GetNumFree(co) > 0) + #define __Pyx_PyFrame_SetLineNumber(frame, lineno) (frame)->f_lineno = (lineno) +#endif +#if !CYTHON_FAST_THREAD_STATE || PY_VERSION_HEX < 0x02070000 + #define __Pyx_PyThreadState_Current PyThreadState_GET() +#elif PY_VERSION_HEX >= 0x03060000 + #define __Pyx_PyThreadState_Current _PyThreadState_UncheckedGet() +#elif PY_VERSION_HEX >= 0x03000000 + #define __Pyx_PyThreadState_Current PyThreadState_GET() +#else + #define __Pyx_PyThreadState_Current _PyThreadState_Current +#endif +#if PY_VERSION_HEX < 0x030700A2 && !defined(PyThread_tss_create) && !defined(Py_tss_NEEDS_INIT) +#include "pythread.h" +#define Py_tss_NEEDS_INIT 0 +typedef int Py_tss_t; +static CYTHON_INLINE int PyThread_tss_create(Py_tss_t *key) { + *key = PyThread_create_key(); + return 0; +} +static CYTHON_INLINE Py_tss_t * PyThread_tss_alloc(void) { + Py_tss_t *key = (Py_tss_t *)PyObject_Malloc(sizeof(Py_tss_t)); + *key = Py_tss_NEEDS_INIT; + return key; +} +static CYTHON_INLINE void PyThread_tss_free(Py_tss_t *key) { + PyObject_Free(key); +} +static CYTHON_INLINE int PyThread_tss_is_created(Py_tss_t *key) { + return *key != Py_tss_NEEDS_INIT; +} +static CYTHON_INLINE void PyThread_tss_delete(Py_tss_t *key) { + PyThread_delete_key(*key); + *key = Py_tss_NEEDS_INIT; +} +static CYTHON_INLINE int PyThread_tss_set(Py_tss_t *key, void *value) { + return PyThread_set_key_value(*key, value); +} +static CYTHON_INLINE void * PyThread_tss_get(Py_tss_t *key) { + return PyThread_get_key_value(*key); +} +#endif +#if CYTHON_COMPILING_IN_CPYTHON || defined(_PyDict_NewPresized) +#define __Pyx_PyDict_NewPresized(n) ((n <= 8) ? PyDict_New() : _PyDict_NewPresized(n)) +#else +#define __Pyx_PyDict_NewPresized(n) PyDict_New() +#endif +#if PY_MAJOR_VERSION >= 3 || CYTHON_FUTURE_DIVISION + #define __Pyx_PyNumber_Divide(x,y) PyNumber_TrueDivide(x,y) + #define __Pyx_PyNumber_InPlaceDivide(x,y) PyNumber_InPlaceTrueDivide(x,y) +#else + #define __Pyx_PyNumber_Divide(x,y) PyNumber_Divide(x,y) + #define __Pyx_PyNumber_InPlaceDivide(x,y) PyNumber_InPlaceDivide(x,y) +#endif +#if CYTHON_COMPILING_IN_CPYTHON && PY_VERSION_HEX >= 0x030500A1 && CYTHON_USE_UNICODE_INTERNALS +#define __Pyx_PyDict_GetItemStr(dict, name) _PyDict_GetItem_KnownHash(dict, name, ((PyASCIIObject *) name)->hash) +#else +#define __Pyx_PyDict_GetItemStr(dict, name) PyDict_GetItem(dict, name) +#endif +#if PY_VERSION_HEX > 0x03030000 && defined(PyUnicode_KIND) + #define CYTHON_PEP393_ENABLED 1 + #define __Pyx_PyUnicode_READY(op) (likely(PyUnicode_IS_READY(op)) ?\ + 0 : _PyUnicode_Ready((PyObject *)(op))) + #define __Pyx_PyUnicode_GET_LENGTH(u) PyUnicode_GET_LENGTH(u) + #define __Pyx_PyUnicode_READ_CHAR(u, i) PyUnicode_READ_CHAR(u, i) + #define __Pyx_PyUnicode_MAX_CHAR_VALUE(u) PyUnicode_MAX_CHAR_VALUE(u) + #define __Pyx_PyUnicode_KIND(u) PyUnicode_KIND(u) + #define __Pyx_PyUnicode_DATA(u) PyUnicode_DATA(u) + #define __Pyx_PyUnicode_READ(k, d, i) PyUnicode_READ(k, d, i) + #define __Pyx_PyUnicode_WRITE(k, d, i, ch) PyUnicode_WRITE(k, d, i, ch) + #define __Pyx_PyUnicode_IS_TRUE(u) (0 != (likely(PyUnicode_IS_READY(u)) ? PyUnicode_GET_LENGTH(u) : PyUnicode_GET_SIZE(u))) +#else + #define CYTHON_PEP393_ENABLED 0 + #define PyUnicode_1BYTE_KIND 1 + #define PyUnicode_2BYTE_KIND 2 + #define PyUnicode_4BYTE_KIND 4 + #define __Pyx_PyUnicode_READY(op) (0) + #define __Pyx_PyUnicode_GET_LENGTH(u) PyUnicode_GET_SIZE(u) + #define __Pyx_PyUnicode_READ_CHAR(u, i) ((Py_UCS4)(PyUnicode_AS_UNICODE(u)[i])) + #define __Pyx_PyUnicode_MAX_CHAR_VALUE(u) ((sizeof(Py_UNICODE) == 2) ? 65535 : 1114111) + #define __Pyx_PyUnicode_KIND(u) (sizeof(Py_UNICODE)) + #define __Pyx_PyUnicode_DATA(u) ((void*)PyUnicode_AS_UNICODE(u)) + #define __Pyx_PyUnicode_READ(k, d, i) ((void)(k), (Py_UCS4)(((Py_UNICODE*)d)[i])) + #define __Pyx_PyUnicode_WRITE(k, d, i, ch) (((void)(k)), ((Py_UNICODE*)d)[i] = ch) + #define __Pyx_PyUnicode_IS_TRUE(u) (0 != PyUnicode_GET_SIZE(u)) +#endif +#if CYTHON_COMPILING_IN_PYPY + #define __Pyx_PyUnicode_Concat(a, b) PyNumber_Add(a, b) + #define __Pyx_PyUnicode_ConcatSafe(a, b) PyNumber_Add(a, b) +#else + #define __Pyx_PyUnicode_Concat(a, b) PyUnicode_Concat(a, b) + #define __Pyx_PyUnicode_ConcatSafe(a, b) ((unlikely((a) == Py_None) || unlikely((b) == Py_None)) ?\ + PyNumber_Add(a, b) : __Pyx_PyUnicode_Concat(a, b)) +#endif +#if CYTHON_COMPILING_IN_PYPY && !defined(PyUnicode_Contains) + #define PyUnicode_Contains(u, s) PySequence_Contains(u, s) +#endif +#if CYTHON_COMPILING_IN_PYPY && !defined(PyByteArray_Check) + #define PyByteArray_Check(obj) PyObject_TypeCheck(obj, &PyByteArray_Type) +#endif +#if CYTHON_COMPILING_IN_PYPY && !defined(PyObject_Format) + #define PyObject_Format(obj, fmt) PyObject_CallMethod(obj, "__format__", "O", fmt) +#endif +#define __Pyx_PyString_FormatSafe(a, b) ((unlikely((a) == Py_None || (PyString_Check(b) && !PyString_CheckExact(b)))) ? PyNumber_Remainder(a, b) : __Pyx_PyString_Format(a, b)) +#define __Pyx_PyUnicode_FormatSafe(a, b) ((unlikely((a) == Py_None || (PyUnicode_Check(b) && !PyUnicode_CheckExact(b)))) ? PyNumber_Remainder(a, b) : PyUnicode_Format(a, b)) +#if PY_MAJOR_VERSION >= 3 + #define __Pyx_PyString_Format(a, b) PyUnicode_Format(a, b) +#else + #define __Pyx_PyString_Format(a, b) PyString_Format(a, b) +#endif +#if PY_MAJOR_VERSION < 3 && !defined(PyObject_ASCII) + #define PyObject_ASCII(o) PyObject_Repr(o) +#endif +#if PY_MAJOR_VERSION >= 3 + #define PyBaseString_Type PyUnicode_Type + #define PyStringObject PyUnicodeObject + #define PyString_Type PyUnicode_Type + #define PyString_Check PyUnicode_Check + #define PyString_CheckExact PyUnicode_CheckExact + #define PyObject_Unicode PyObject_Str +#endif +#if PY_MAJOR_VERSION >= 3 + #define __Pyx_PyBaseString_Check(obj) PyUnicode_Check(obj) + #define __Pyx_PyBaseString_CheckExact(obj) PyUnicode_CheckExact(obj) +#else + #define __Pyx_PyBaseString_Check(obj) (PyString_Check(obj) || PyUnicode_Check(obj)) + #define __Pyx_PyBaseString_CheckExact(obj) (PyString_CheckExact(obj) || PyUnicode_CheckExact(obj)) +#endif +#ifndef PySet_CheckExact + #define PySet_CheckExact(obj) (Py_TYPE(obj) == &PySet_Type) +#endif +#if CYTHON_ASSUME_SAFE_MACROS + #define __Pyx_PySequence_SIZE(seq) Py_SIZE(seq) +#else + #define __Pyx_PySequence_SIZE(seq) PySequence_Size(seq) +#endif +#if PY_MAJOR_VERSION >= 3 + #define PyIntObject PyLongObject + #define PyInt_Type PyLong_Type + #define PyInt_Check(op) PyLong_Check(op) + #define PyInt_CheckExact(op) PyLong_CheckExact(op) + #define PyInt_FromString PyLong_FromString + #define PyInt_FromUnicode PyLong_FromUnicode + #define PyInt_FromLong PyLong_FromLong + #define PyInt_FromSize_t PyLong_FromSize_t + #define PyInt_FromSsize_t PyLong_FromSsize_t + #define PyInt_AsLong PyLong_AsLong + #define PyInt_AS_LONG PyLong_AS_LONG + #define PyInt_AsSsize_t PyLong_AsSsize_t + #define PyInt_AsUnsignedLongMask PyLong_AsUnsignedLongMask + #define PyInt_AsUnsignedLongLongMask PyLong_AsUnsignedLongLongMask + #define PyNumber_Int PyNumber_Long +#endif +#if PY_MAJOR_VERSION >= 3 + #define PyBoolObject PyLongObject +#endif +#if PY_MAJOR_VERSION >= 3 && CYTHON_COMPILING_IN_PYPY + #ifndef PyUnicode_InternFromString + #define PyUnicode_InternFromString(s) PyUnicode_FromString(s) + #endif +#endif +#if PY_VERSION_HEX < 0x030200A4 + typedef long Py_hash_t; + #define __Pyx_PyInt_FromHash_t PyInt_FromLong + #define __Pyx_PyInt_AsHash_t PyInt_AsLong +#else + #define __Pyx_PyInt_FromHash_t PyInt_FromSsize_t + #define __Pyx_PyInt_AsHash_t PyInt_AsSsize_t +#endif +#if PY_MAJOR_VERSION >= 3 + #define __Pyx_PyMethod_New(func, self, klass) ((self) ? PyMethod_New(func, self) : (Py_INCREF(func), func)) +#else + #define __Pyx_PyMethod_New(func, self, klass) PyMethod_New(func, self, klass) +#endif +#if CYTHON_USE_ASYNC_SLOTS + #if PY_VERSION_HEX >= 0x030500B1 + #define __Pyx_PyAsyncMethodsStruct PyAsyncMethods + #define __Pyx_PyType_AsAsync(obj) (Py_TYPE(obj)->tp_as_async) + #else + #define __Pyx_PyType_AsAsync(obj) ((__Pyx_PyAsyncMethodsStruct*) (Py_TYPE(obj)->tp_reserved)) + #endif +#else + #define __Pyx_PyType_AsAsync(obj) NULL +#endif +#ifndef __Pyx_PyAsyncMethodsStruct + typedef struct { + unaryfunc am_await; + unaryfunc am_aiter; + unaryfunc am_anext; + } __Pyx_PyAsyncMethodsStruct; +#endif + +#if defined(WIN32) || defined(MS_WINDOWS) + #define _USE_MATH_DEFINES +#endif +#include +#ifdef NAN +#define __PYX_NAN() ((float) NAN) +#else +static CYTHON_INLINE float __PYX_NAN() { + float value; + memset(&value, 0xFF, sizeof(value)); + return value; +} +#endif +#if defined(__CYGWIN__) && defined(_LDBL_EQ_DBL) +#define __Pyx_truncl trunc +#else +#define __Pyx_truncl truncl +#endif + + +#define __PYX_ERR(f_index, lineno, Ln_error) \ +{ \ + __pyx_filename = __pyx_f[f_index]; __pyx_lineno = lineno; __pyx_clineno = __LINE__; goto Ln_error; \ +} + +#ifndef __PYX_EXTERN_C + #ifdef __cplusplus + #define __PYX_EXTERN_C extern "C" + #else + #define __PYX_EXTERN_C extern + #endif +#endif + +#define __PYX_HAVE__pooling_inner +#define __PYX_HAVE_API__pooling_inner +/* Early includes */ +#include +#include +#include "numpy/arrayobject.h" +#include "numpy/ufuncobject.h" +#include "voidptr.h" +#ifdef _OPENMP +#include +#endif /* _OPENMP */ + +#if defined(PYREX_WITHOUT_ASSERTIONS) && !defined(CYTHON_WITHOUT_ASSERTIONS) +#define CYTHON_WITHOUT_ASSERTIONS +#endif + +typedef struct {PyObject **p; const char *s; const Py_ssize_t n; const char* encoding; + const char is_unicode; const char is_str; const char intern; } __Pyx_StringTabEntry; + +#define __PYX_DEFAULT_STRING_ENCODING_IS_ASCII 0 +#define __PYX_DEFAULT_STRING_ENCODING_IS_UTF8 0 +#define __PYX_DEFAULT_STRING_ENCODING_IS_DEFAULT (PY_MAJOR_VERSION >= 3 && __PYX_DEFAULT_STRING_ENCODING_IS_UTF8) +#define __PYX_DEFAULT_STRING_ENCODING "" +#define __Pyx_PyObject_FromString __Pyx_PyBytes_FromString +#define __Pyx_PyObject_FromStringAndSize __Pyx_PyBytes_FromStringAndSize +#define __Pyx_uchar_cast(c) ((unsigned char)c) +#define __Pyx_long_cast(x) ((long)x) +#define __Pyx_fits_Py_ssize_t(v, type, is_signed) (\ + (sizeof(type) < sizeof(Py_ssize_t)) ||\ + (sizeof(type) > sizeof(Py_ssize_t) &&\ + likely(v < (type)PY_SSIZE_T_MAX ||\ + v == (type)PY_SSIZE_T_MAX) &&\ + (!is_signed || likely(v > (type)PY_SSIZE_T_MIN ||\ + v == (type)PY_SSIZE_T_MIN))) ||\ + (sizeof(type) == sizeof(Py_ssize_t) &&\ + (is_signed || likely(v < (type)PY_SSIZE_T_MAX ||\ + v == (type)PY_SSIZE_T_MAX))) ) +static CYTHON_INLINE int __Pyx_is_valid_index(Py_ssize_t i, Py_ssize_t limit) { + return (size_t) i < (size_t) limit; +} +#if defined (__cplusplus) && __cplusplus >= 201103L + #include + #define __Pyx_sst_abs(value) std::abs(value) +#elif SIZEOF_INT >= SIZEOF_SIZE_T + #define __Pyx_sst_abs(value) abs(value) +#elif SIZEOF_LONG >= SIZEOF_SIZE_T + #define __Pyx_sst_abs(value) labs(value) +#elif defined (_MSC_VER) + #define __Pyx_sst_abs(value) ((Py_ssize_t)_abs64(value)) +#elif defined (__STDC_VERSION__) && __STDC_VERSION__ >= 199901L + #define __Pyx_sst_abs(value) llabs(value) +#elif defined (__GNUC__) + #define __Pyx_sst_abs(value) __builtin_llabs(value) +#else + #define __Pyx_sst_abs(value) ((value<0) ? -value : value) +#endif +static CYTHON_INLINE const char* __Pyx_PyObject_AsString(PyObject*); +static CYTHON_INLINE const char* __Pyx_PyObject_AsStringAndSize(PyObject*, Py_ssize_t* length); +#define __Pyx_PyByteArray_FromString(s) PyByteArray_FromStringAndSize((const char*)s, strlen((const char*)s)) +#define __Pyx_PyByteArray_FromStringAndSize(s, l) PyByteArray_FromStringAndSize((const char*)s, l) +#define __Pyx_PyBytes_FromString PyBytes_FromString +#define __Pyx_PyBytes_FromStringAndSize PyBytes_FromStringAndSize +static CYTHON_INLINE PyObject* __Pyx_PyUnicode_FromString(const char*); +#if PY_MAJOR_VERSION < 3 + #define __Pyx_PyStr_FromString __Pyx_PyBytes_FromString + #define __Pyx_PyStr_FromStringAndSize __Pyx_PyBytes_FromStringAndSize +#else + #define __Pyx_PyStr_FromString __Pyx_PyUnicode_FromString + #define __Pyx_PyStr_FromStringAndSize __Pyx_PyUnicode_FromStringAndSize +#endif +#define __Pyx_PyBytes_AsWritableString(s) ((char*) PyBytes_AS_STRING(s)) +#define __Pyx_PyBytes_AsWritableSString(s) ((signed char*) PyBytes_AS_STRING(s)) +#define __Pyx_PyBytes_AsWritableUString(s) ((unsigned char*) PyBytes_AS_STRING(s)) +#define __Pyx_PyBytes_AsString(s) ((const char*) PyBytes_AS_STRING(s)) +#define __Pyx_PyBytes_AsSString(s) ((const signed char*) PyBytes_AS_STRING(s)) +#define __Pyx_PyBytes_AsUString(s) ((const unsigned char*) PyBytes_AS_STRING(s)) +#define __Pyx_PyObject_AsWritableString(s) ((char*) __Pyx_PyObject_AsString(s)) +#define __Pyx_PyObject_AsWritableSString(s) ((signed char*) __Pyx_PyObject_AsString(s)) +#define __Pyx_PyObject_AsWritableUString(s) ((unsigned char*) __Pyx_PyObject_AsString(s)) +#define __Pyx_PyObject_AsSString(s) ((const signed char*) __Pyx_PyObject_AsString(s)) +#define __Pyx_PyObject_AsUString(s) ((const unsigned char*) __Pyx_PyObject_AsString(s)) +#define __Pyx_PyObject_FromCString(s) __Pyx_PyObject_FromString((const char*)s) +#define __Pyx_PyBytes_FromCString(s) __Pyx_PyBytes_FromString((const char*)s) +#define __Pyx_PyByteArray_FromCString(s) __Pyx_PyByteArray_FromString((const char*)s) +#define __Pyx_PyStr_FromCString(s) __Pyx_PyStr_FromString((const char*)s) +#define __Pyx_PyUnicode_FromCString(s) __Pyx_PyUnicode_FromString((const char*)s) +static CYTHON_INLINE size_t __Pyx_Py_UNICODE_strlen(const Py_UNICODE *u) { + const Py_UNICODE *u_end = u; + while (*u_end++) ; + return (size_t)(u_end - u - 1); +} +#define __Pyx_PyUnicode_FromUnicode(u) PyUnicode_FromUnicode(u, __Pyx_Py_UNICODE_strlen(u)) +#define __Pyx_PyUnicode_FromUnicodeAndLength PyUnicode_FromUnicode +#define __Pyx_PyUnicode_AsUnicode PyUnicode_AsUnicode +#define __Pyx_NewRef(obj) (Py_INCREF(obj), obj) +#define __Pyx_Owned_Py_None(b) __Pyx_NewRef(Py_None) +static CYTHON_INLINE PyObject * __Pyx_PyBool_FromLong(long b); +static CYTHON_INLINE int __Pyx_PyObject_IsTrue(PyObject*); +static CYTHON_INLINE int __Pyx_PyObject_IsTrueAndDecref(PyObject*); +static CYTHON_INLINE PyObject* __Pyx_PyNumber_IntOrLong(PyObject* x); +#define __Pyx_PySequence_Tuple(obj)\ + (likely(PyTuple_CheckExact(obj)) ? __Pyx_NewRef(obj) : PySequence_Tuple(obj)) +static CYTHON_INLINE Py_ssize_t __Pyx_PyIndex_AsSsize_t(PyObject*); +static CYTHON_INLINE PyObject * __Pyx_PyInt_FromSize_t(size_t); +#if CYTHON_ASSUME_SAFE_MACROS +#define __pyx_PyFloat_AsDouble(x) (PyFloat_CheckExact(x) ? PyFloat_AS_DOUBLE(x) : PyFloat_AsDouble(x)) +#else +#define __pyx_PyFloat_AsDouble(x) PyFloat_AsDouble(x) +#endif +#define __pyx_PyFloat_AsFloat(x) ((float) __pyx_PyFloat_AsDouble(x)) +#if PY_MAJOR_VERSION >= 3 +#define __Pyx_PyNumber_Int(x) (PyLong_CheckExact(x) ? __Pyx_NewRef(x) : PyNumber_Long(x)) +#else +#define __Pyx_PyNumber_Int(x) (PyInt_CheckExact(x) ? __Pyx_NewRef(x) : PyNumber_Int(x)) +#endif +#define __Pyx_PyNumber_Float(x) (PyFloat_CheckExact(x) ? __Pyx_NewRef(x) : PyNumber_Float(x)) +#if PY_MAJOR_VERSION < 3 && __PYX_DEFAULT_STRING_ENCODING_IS_ASCII +static int __Pyx_sys_getdefaultencoding_not_ascii; +static int __Pyx_init_sys_getdefaultencoding_params(void) { + PyObject* sys; + PyObject* default_encoding = NULL; + PyObject* ascii_chars_u = NULL; + PyObject* ascii_chars_b = NULL; + const char* default_encoding_c; + sys = PyImport_ImportModule("sys"); + if (!sys) goto bad; + default_encoding = PyObject_CallMethod(sys, (char*) "getdefaultencoding", NULL); + Py_DECREF(sys); + if (!default_encoding) goto bad; + default_encoding_c = PyBytes_AsString(default_encoding); + if (!default_encoding_c) goto bad; + if (strcmp(default_encoding_c, "ascii") == 0) { + __Pyx_sys_getdefaultencoding_not_ascii = 0; + } else { + char ascii_chars[128]; + int c; + for (c = 0; c < 128; c++) { + ascii_chars[c] = c; + } + __Pyx_sys_getdefaultencoding_not_ascii = 1; + ascii_chars_u = PyUnicode_DecodeASCII(ascii_chars, 128, NULL); + if (!ascii_chars_u) goto bad; + ascii_chars_b = PyUnicode_AsEncodedString(ascii_chars_u, default_encoding_c, NULL); + if (!ascii_chars_b || !PyBytes_Check(ascii_chars_b) || memcmp(ascii_chars, PyBytes_AS_STRING(ascii_chars_b), 128) != 0) { + PyErr_Format( + PyExc_ValueError, + "This module compiled with c_string_encoding=ascii, but default encoding '%.200s' is not a superset of ascii.", + default_encoding_c); + goto bad; + } + Py_DECREF(ascii_chars_u); + Py_DECREF(ascii_chars_b); + } + Py_DECREF(default_encoding); + return 0; +bad: + Py_XDECREF(default_encoding); + Py_XDECREF(ascii_chars_u); + Py_XDECREF(ascii_chars_b); + return -1; +} +#endif +#if __PYX_DEFAULT_STRING_ENCODING_IS_DEFAULT && PY_MAJOR_VERSION >= 3 +#define __Pyx_PyUnicode_FromStringAndSize(c_str, size) PyUnicode_DecodeUTF8(c_str, size, NULL) +#else +#define __Pyx_PyUnicode_FromStringAndSize(c_str, size) PyUnicode_Decode(c_str, size, __PYX_DEFAULT_STRING_ENCODING, NULL) +#if __PYX_DEFAULT_STRING_ENCODING_IS_DEFAULT +static char* __PYX_DEFAULT_STRING_ENCODING; +static int __Pyx_init_sys_getdefaultencoding_params(void) { + PyObject* sys; + PyObject* default_encoding = NULL; + char* default_encoding_c; + sys = PyImport_ImportModule("sys"); + if (!sys) goto bad; + default_encoding = PyObject_CallMethod(sys, (char*) (const char*) "getdefaultencoding", NULL); + Py_DECREF(sys); + if (!default_encoding) goto bad; + default_encoding_c = PyBytes_AsString(default_encoding); + if (!default_encoding_c) goto bad; + __PYX_DEFAULT_STRING_ENCODING = (char*) malloc(strlen(default_encoding_c) + 1); + if (!__PYX_DEFAULT_STRING_ENCODING) goto bad; + strcpy(__PYX_DEFAULT_STRING_ENCODING, default_encoding_c); + Py_DECREF(default_encoding); + return 0; +bad: + Py_XDECREF(default_encoding); + return -1; +} +#endif +#endif + + +/* Test for GCC > 2.95 */ +#if defined(__GNUC__) && (__GNUC__ > 2 || (__GNUC__ == 2 && (__GNUC_MINOR__ > 95))) + #define likely(x) __builtin_expect(!!(x), 1) + #define unlikely(x) __builtin_expect(!!(x), 0) +#else /* !__GNUC__ or GCC < 2.95 */ + #define likely(x) (x) + #define unlikely(x) (x) +#endif /* __GNUC__ */ +static CYTHON_INLINE void __Pyx_pretend_to_initialize(void* ptr) { (void)ptr; } + +static PyObject *__pyx_m = NULL; +static PyObject *__pyx_d; +static PyObject *__pyx_b; +static PyObject *__pyx_cython_runtime = NULL; +static PyObject *__pyx_empty_tuple; +static PyObject *__pyx_empty_bytes; +static PyObject *__pyx_empty_unicode; +static int __pyx_lineno; +static int __pyx_clineno = 0; +static const char * __pyx_cfilenm= __FILE__; +static const char *__pyx_filename; + +/* Header.proto */ +#if !defined(CYTHON_CCOMPLEX) + #if defined(__cplusplus) + #define CYTHON_CCOMPLEX 1 + #elif defined(_Complex_I) + #define CYTHON_CCOMPLEX 1 + #else + #define CYTHON_CCOMPLEX 0 + #endif +#endif +#if CYTHON_CCOMPLEX + #ifdef __cplusplus + #include + #else + #include + #endif +#endif +#if CYTHON_CCOMPLEX && !defined(__cplusplus) && defined(__sun__) && defined(__GNUC__) + #undef _Complex_I + #define _Complex_I 1.0fj +#endif + + +static const char *__pyx_f[] = { + "pooling_inner.pyx", + "__init__.pxd", + "type.pxd", +}; +/* NoFastGil.proto */ +#define __Pyx_PyGILState_Ensure PyGILState_Ensure +#define __Pyx_PyGILState_Release PyGILState_Release +#define __Pyx_FastGIL_Remember() +#define __Pyx_FastGIL_Forget() +#define __Pyx_FastGilFuncInit() + +/* ForceInitThreads.proto */ +#ifndef __PYX_FORCE_INIT_THREADS + #define __PYX_FORCE_INIT_THREADS 0 +#endif + + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":776 + * # in Cython to enable them only on the right systems. + * + * ctypedef npy_int8 int8_t # <<<<<<<<<<<<<< + * ctypedef npy_int16 int16_t + * ctypedef npy_int32 int32_t + */ +typedef npy_int8 __pyx_t_5numpy_int8_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":777 + * + * ctypedef npy_int8 int8_t + * ctypedef npy_int16 int16_t # <<<<<<<<<<<<<< + * ctypedef npy_int32 int32_t + * ctypedef npy_int64 int64_t + */ +typedef npy_int16 __pyx_t_5numpy_int16_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":778 + * ctypedef npy_int8 int8_t + * ctypedef npy_int16 int16_t + * ctypedef npy_int32 int32_t # <<<<<<<<<<<<<< + * ctypedef npy_int64 int64_t + * #ctypedef npy_int96 int96_t + */ +typedef npy_int32 __pyx_t_5numpy_int32_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":779 + * ctypedef npy_int16 int16_t + * ctypedef npy_int32 int32_t + * ctypedef npy_int64 int64_t # <<<<<<<<<<<<<< + * #ctypedef npy_int96 int96_t + * #ctypedef npy_int128 int128_t + */ +typedef npy_int64 __pyx_t_5numpy_int64_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":783 + * #ctypedef npy_int128 int128_t + * + * ctypedef npy_uint8 uint8_t # <<<<<<<<<<<<<< + * ctypedef npy_uint16 uint16_t + * ctypedef npy_uint32 uint32_t + */ +typedef npy_uint8 __pyx_t_5numpy_uint8_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":784 + * + * ctypedef npy_uint8 uint8_t + * ctypedef npy_uint16 uint16_t # <<<<<<<<<<<<<< + * ctypedef npy_uint32 uint32_t + * ctypedef npy_uint64 uint64_t + */ +typedef npy_uint16 __pyx_t_5numpy_uint16_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":785 + * ctypedef npy_uint8 uint8_t + * ctypedef npy_uint16 uint16_t + * ctypedef npy_uint32 uint32_t # <<<<<<<<<<<<<< + * ctypedef npy_uint64 uint64_t + * #ctypedef npy_uint96 uint96_t + */ +typedef npy_uint32 __pyx_t_5numpy_uint32_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":786 + * ctypedef npy_uint16 uint16_t + * ctypedef npy_uint32 uint32_t + * ctypedef npy_uint64 uint64_t # <<<<<<<<<<<<<< + * #ctypedef npy_uint96 uint96_t + * #ctypedef npy_uint128 uint128_t + */ +typedef npy_uint64 __pyx_t_5numpy_uint64_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":790 + * #ctypedef npy_uint128 uint128_t + * + * ctypedef npy_float32 float32_t # <<<<<<<<<<<<<< + * ctypedef npy_float64 float64_t + * #ctypedef npy_float80 float80_t + */ +typedef npy_float32 __pyx_t_5numpy_float32_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":791 + * + * ctypedef npy_float32 float32_t + * ctypedef npy_float64 float64_t # <<<<<<<<<<<<<< + * #ctypedef npy_float80 float80_t + * #ctypedef npy_float128 float128_t + */ +typedef npy_float64 __pyx_t_5numpy_float64_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":800 + * # The int types are mapped a bit surprising -- + * # numpy.int corresponds to 'l' and numpy.long to 'q' + * ctypedef npy_long int_t # <<<<<<<<<<<<<< + * ctypedef npy_longlong long_t + * ctypedef npy_longlong longlong_t + */ +typedef npy_long __pyx_t_5numpy_int_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":801 + * # numpy.int corresponds to 'l' and numpy.long to 'q' + * ctypedef npy_long int_t + * ctypedef npy_longlong long_t # <<<<<<<<<<<<<< + * ctypedef npy_longlong longlong_t + * + */ +typedef npy_longlong __pyx_t_5numpy_long_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":802 + * ctypedef npy_long int_t + * ctypedef npy_longlong long_t + * ctypedef npy_longlong longlong_t # <<<<<<<<<<<<<< + * + * ctypedef npy_ulong uint_t + */ +typedef npy_longlong __pyx_t_5numpy_longlong_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":804 + * ctypedef npy_longlong longlong_t + * + * ctypedef npy_ulong uint_t # <<<<<<<<<<<<<< + * ctypedef npy_ulonglong ulong_t + * ctypedef npy_ulonglong ulonglong_t + */ +typedef npy_ulong __pyx_t_5numpy_uint_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":805 + * + * ctypedef npy_ulong uint_t + * ctypedef npy_ulonglong ulong_t # <<<<<<<<<<<<<< + * ctypedef npy_ulonglong ulonglong_t + * + */ +typedef npy_ulonglong __pyx_t_5numpy_ulong_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":806 + * ctypedef npy_ulong uint_t + * ctypedef npy_ulonglong ulong_t + * ctypedef npy_ulonglong ulonglong_t # <<<<<<<<<<<<<< + * + * ctypedef npy_intp intp_t + */ +typedef npy_ulonglong __pyx_t_5numpy_ulonglong_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":808 + * ctypedef npy_ulonglong ulonglong_t + * + * ctypedef npy_intp intp_t # <<<<<<<<<<<<<< + * ctypedef npy_uintp uintp_t + * + */ +typedef npy_intp __pyx_t_5numpy_intp_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":809 + * + * ctypedef npy_intp intp_t + * ctypedef npy_uintp uintp_t # <<<<<<<<<<<<<< + * + * ctypedef npy_double float_t + */ +typedef npy_uintp __pyx_t_5numpy_uintp_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":811 + * ctypedef npy_uintp uintp_t + * + * ctypedef npy_double float_t # <<<<<<<<<<<<<< + * ctypedef npy_double double_t + * ctypedef npy_longdouble longdouble_t + */ +typedef npy_double __pyx_t_5numpy_float_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":812 + * + * ctypedef npy_double float_t + * ctypedef npy_double double_t # <<<<<<<<<<<<<< + * ctypedef npy_longdouble longdouble_t + * + */ +typedef npy_double __pyx_t_5numpy_double_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":813 + * ctypedef npy_double float_t + * ctypedef npy_double double_t + * ctypedef npy_longdouble longdouble_t # <<<<<<<<<<<<<< + * + * ctypedef npy_cfloat cfloat_t + */ +typedef npy_longdouble __pyx_t_5numpy_longdouble_t; + +/* "average_inner.pxd":15 + * void* PyCObject_AsVoidPtr(object obj) + * + * ctypedef np.float32_t REAL_t # <<<<<<<<<<<<<< + * ctypedef np.uint32_t uINT_t + * + */ +typedef __pyx_t_5numpy_float32_t __pyx_t_13average_inner_REAL_t; + +/* "average_inner.pxd":16 + * + * ctypedef np.float32_t REAL_t + * ctypedef np.uint32_t uINT_t # <<<<<<<<<<<<<< + * + * # BLAS routine signatures + */ +typedef __pyx_t_5numpy_uint32_t __pyx_t_13average_inner_uINT_t; +/* Declarations.proto */ +#if CYTHON_CCOMPLEX + #ifdef __cplusplus + typedef ::std::complex< float > __pyx_t_float_complex; + #else + typedef float _Complex __pyx_t_float_complex; + #endif +#else + typedef struct { float real, imag; } __pyx_t_float_complex; +#endif +static CYTHON_INLINE __pyx_t_float_complex __pyx_t_float_complex_from_parts(float, float); + +/* Declarations.proto */ +#if CYTHON_CCOMPLEX + #ifdef __cplusplus + typedef ::std::complex< double > __pyx_t_double_complex; + #else + typedef double _Complex __pyx_t_double_complex; + #endif +#else + typedef struct { double real, imag; } __pyx_t_double_complex; +#endif +static CYTHON_INLINE __pyx_t_double_complex __pyx_t_double_complex_from_parts(double, double); + + +/*--- Type declarations ---*/ + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":815 + * ctypedef npy_longdouble longdouble_t + * + * ctypedef npy_cfloat cfloat_t # <<<<<<<<<<<<<< + * ctypedef npy_cdouble cdouble_t + * ctypedef npy_clongdouble clongdouble_t + */ +typedef npy_cfloat __pyx_t_5numpy_cfloat_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":816 + * + * ctypedef npy_cfloat cfloat_t + * ctypedef npy_cdouble cdouble_t # <<<<<<<<<<<<<< + * ctypedef npy_clongdouble clongdouble_t + * + */ +typedef npy_cdouble __pyx_t_5numpy_cdouble_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":817 + * ctypedef npy_cfloat cfloat_t + * ctypedef npy_cdouble cdouble_t + * ctypedef npy_clongdouble clongdouble_t # <<<<<<<<<<<<<< + * + * ctypedef npy_cdouble complex_t + */ +typedef npy_clongdouble __pyx_t_5numpy_clongdouble_t; + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":819 + * ctypedef npy_clongdouble clongdouble_t + * + * ctypedef npy_cdouble complex_t # <<<<<<<<<<<<<< + * + * cdef inline object PyArray_MultiIterNew1(a): + */ +typedef npy_cdouble __pyx_t_5numpy_complex_t; +struct __pyx_t_13average_inner_BaseSentenceVecsConfig; +struct __pyx_t_13average_inner_FTSentenceVecsConfig; + +/* "average_inner.pxd":19 + * + * # BLAS routine signatures + * ctypedef void (*saxpy_ptr) (const int *N, const float *alpha, const float *X, const int *incX, float *Y, const int *incY) nogil # <<<<<<<<<<<<<< + * ctypedef void (*sscal_ptr) (const int *N, const float *alpha, const float *X, const int *incX) nogil + * + */ +typedef void (*__pyx_t_13average_inner_saxpy_ptr)(int const *, float const *, float const *, int const *, float *, int const *); + +/* "average_inner.pxd":20 + * # BLAS routine signatures + * ctypedef void (*saxpy_ptr) (const int *N, const float *alpha, const float *X, const int *incX, float *Y, const int *incY) nogil + * ctypedef void (*sscal_ptr) (const int *N, const float *alpha, const float *X, const int *incX) nogil # <<<<<<<<<<<<<< + * + * cdef saxpy_ptr saxpy + */ +typedef void (*__pyx_t_13average_inner_sscal_ptr)(int const *, float const *, float const *, int const *); + +/* "average_inner.pxd":34 + * DEF MAX_NGRAMS = 40 + * + * cdef struct BaseSentenceVecsConfig: # <<<<<<<<<<<<<< + * int size, workers + * + */ +struct __pyx_t_13average_inner_BaseSentenceVecsConfig { + int size; + int workers; + __pyx_t_13average_inner_REAL_t *mem; + __pyx_t_13average_inner_REAL_t *mem2; + __pyx_t_13average_inner_REAL_t *word_vectors; + __pyx_t_13average_inner_REAL_t *word_weights; + __pyx_t_13average_inner_REAL_t *sentence_vectors; + __pyx_t_13average_inner_uINT_t word_indices[0x2710]; + __pyx_t_13average_inner_uINT_t sent_adresses[0x2710]; + __pyx_t_13average_inner_uINT_t sentence_boundary[(0x2710 + 1)]; +}; + +/* "average_inner.pxd":48 + * uINT_t sentence_boundary[MAX_WORDS + 1] + * + * cdef struct FTSentenceVecsConfig: # <<<<<<<<<<<<<< + * int size, workers, min_n, max_n, bucket + * + */ +struct __pyx_t_13average_inner_FTSentenceVecsConfig { + int size; + int workers; + int min_n; + int max_n; + int bucket; + __pyx_t_13average_inner_REAL_t oov_weight; + __pyx_t_13average_inner_REAL_t *mem; + __pyx_t_13average_inner_REAL_t *mem2; + __pyx_t_13average_inner_REAL_t *word_vectors; + __pyx_t_13average_inner_REAL_t *ngram_vectors; + __pyx_t_13average_inner_REAL_t *word_weights; + __pyx_t_13average_inner_REAL_t *sentence_vectors; + __pyx_t_13average_inner_uINT_t word_indices[0x2710]; + __pyx_t_13average_inner_uINT_t sent_adresses[0x2710]; + __pyx_t_13average_inner_uINT_t sentence_boundary[(0x2710 + 1)]; + __pyx_t_13average_inner_uINT_t subwords_idx_len[0x2710]; + __pyx_t_13average_inner_uINT_t *subwords_idx; +}; + +/* --- Runtime support code (head) --- */ +/* Refnanny.proto */ +#ifndef CYTHON_REFNANNY + #define CYTHON_REFNANNY 0 +#endif +#if CYTHON_REFNANNY + typedef struct { + void (*INCREF)(void*, PyObject*, int); + void (*DECREF)(void*, PyObject*, int); + void (*GOTREF)(void*, PyObject*, int); + void (*GIVEREF)(void*, PyObject*, int); + void* (*SetupContext)(const char*, int, const char*); + void (*FinishContext)(void**); + } __Pyx_RefNannyAPIStruct; + static __Pyx_RefNannyAPIStruct *__Pyx_RefNanny = NULL; + static __Pyx_RefNannyAPIStruct *__Pyx_RefNannyImportAPI(const char *modname); + #define __Pyx_RefNannyDeclarations void *__pyx_refnanny = NULL; +#ifdef WITH_THREAD + #define __Pyx_RefNannySetupContext(name, acquire_gil)\ + if (acquire_gil) {\ + PyGILState_STATE __pyx_gilstate_save = PyGILState_Ensure();\ + __pyx_refnanny = __Pyx_RefNanny->SetupContext((name), __LINE__, __FILE__);\ + PyGILState_Release(__pyx_gilstate_save);\ + } else {\ + __pyx_refnanny = __Pyx_RefNanny->SetupContext((name), __LINE__, __FILE__);\ + } +#else + #define __Pyx_RefNannySetupContext(name, acquire_gil)\ + __pyx_refnanny = __Pyx_RefNanny->SetupContext((name), __LINE__, __FILE__) +#endif + #define __Pyx_RefNannyFinishContext()\ + __Pyx_RefNanny->FinishContext(&__pyx_refnanny) + #define __Pyx_INCREF(r) __Pyx_RefNanny->INCREF(__pyx_refnanny, (PyObject *)(r), __LINE__) + #define __Pyx_DECREF(r) __Pyx_RefNanny->DECREF(__pyx_refnanny, (PyObject *)(r), __LINE__) + #define __Pyx_GOTREF(r) __Pyx_RefNanny->GOTREF(__pyx_refnanny, (PyObject *)(r), __LINE__) + #define __Pyx_GIVEREF(r) __Pyx_RefNanny->GIVEREF(__pyx_refnanny, (PyObject *)(r), __LINE__) + #define __Pyx_XINCREF(r) do { if((r) != NULL) {__Pyx_INCREF(r); }} while(0) + #define __Pyx_XDECREF(r) do { if((r) != NULL) {__Pyx_DECREF(r); }} while(0) + #define __Pyx_XGOTREF(r) do { if((r) != NULL) {__Pyx_GOTREF(r); }} while(0) + #define __Pyx_XGIVEREF(r) do { if((r) != NULL) {__Pyx_GIVEREF(r);}} while(0) +#else + #define __Pyx_RefNannyDeclarations + #define __Pyx_RefNannySetupContext(name, acquire_gil) + #define __Pyx_RefNannyFinishContext() + #define __Pyx_INCREF(r) Py_INCREF(r) + #define __Pyx_DECREF(r) Py_DECREF(r) + #define __Pyx_GOTREF(r) + #define __Pyx_GIVEREF(r) + #define __Pyx_XINCREF(r) Py_XINCREF(r) + #define __Pyx_XDECREF(r) Py_XDECREF(r) + #define __Pyx_XGOTREF(r) + #define __Pyx_XGIVEREF(r) +#endif +#define __Pyx_XDECREF_SET(r, v) do {\ + PyObject *tmp = (PyObject *) r;\ + r = v; __Pyx_XDECREF(tmp);\ + } while (0) +#define __Pyx_DECREF_SET(r, v) do {\ + PyObject *tmp = (PyObject *) r;\ + r = v; __Pyx_DECREF(tmp);\ + } while (0) +#define __Pyx_CLEAR(r) do { PyObject* tmp = ((PyObject*)(r)); r = NULL; __Pyx_DECREF(tmp);} while(0) +#define __Pyx_XCLEAR(r) do { if((r) != NULL) {PyObject* tmp = ((PyObject*)(r)); r = NULL; __Pyx_DECREF(tmp);}} while(0) + +/* PyObjectGetAttrStr.proto */ +#if CYTHON_USE_TYPE_SLOTS +static CYTHON_INLINE PyObject* __Pyx_PyObject_GetAttrStr(PyObject* obj, PyObject* attr_name); +#else +#define __Pyx_PyObject_GetAttrStr(o,n) PyObject_GetAttr(o,n) +#endif + +/* GetBuiltinName.proto */ +static PyObject *__Pyx_GetBuiltinName(PyObject *name); + +/* RaiseArgTupleInvalid.proto */ +static void __Pyx_RaiseArgtupleInvalid(const char* func_name, int exact, + Py_ssize_t num_min, Py_ssize_t num_max, Py_ssize_t num_found); + +/* RaiseDoubleKeywords.proto */ +static void __Pyx_RaiseDoubleKeywordsError(const char* func_name, PyObject* kw_name); + +/* ParseKeywords.proto */ +static int __Pyx_ParseOptionalKeywords(PyObject *kwds, PyObject **argnames[],\ + PyObject *kwds2, PyObject *values[], Py_ssize_t num_pos_args,\ + const char* function_name); + +/* RaiseTooManyValuesToUnpack.proto */ +static CYTHON_INLINE void __Pyx_RaiseTooManyValuesError(Py_ssize_t expected); + +/* RaiseNeedMoreValuesToUnpack.proto */ +static CYTHON_INLINE void __Pyx_RaiseNeedMoreValuesError(Py_ssize_t index); + +/* IterFinish.proto */ +static CYTHON_INLINE int __Pyx_IterFinish(void); + +/* UnpackItemEndCheck.proto */ +static int __Pyx_IternextUnpackEndCheck(PyObject *retval, Py_ssize_t expected); + +/* PyObjectCall.proto */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE PyObject* __Pyx_PyObject_Call(PyObject *func, PyObject *arg, PyObject *kw); +#else +#define __Pyx_PyObject_Call(func, arg, kw) PyObject_Call(func, arg, kw) +#endif + +/* PyThreadStateGet.proto */ +#if CYTHON_FAST_THREAD_STATE +#define __Pyx_PyThreadState_declare PyThreadState *__pyx_tstate; +#define __Pyx_PyThreadState_assign __pyx_tstate = __Pyx_PyThreadState_Current; +#define __Pyx_PyErr_Occurred() __pyx_tstate->curexc_type +#else +#define __Pyx_PyThreadState_declare +#define __Pyx_PyThreadState_assign +#define __Pyx_PyErr_Occurred() PyErr_Occurred() +#endif + +/* PyErrFetchRestore.proto */ +#if CYTHON_FAST_THREAD_STATE +#define __Pyx_PyErr_Clear() __Pyx_ErrRestore(NULL, NULL, NULL) +#define __Pyx_ErrRestoreWithState(type, value, tb) __Pyx_ErrRestoreInState(PyThreadState_GET(), type, value, tb) +#define __Pyx_ErrFetchWithState(type, value, tb) __Pyx_ErrFetchInState(PyThreadState_GET(), type, value, tb) +#define __Pyx_ErrRestore(type, value, tb) __Pyx_ErrRestoreInState(__pyx_tstate, type, value, tb) +#define __Pyx_ErrFetch(type, value, tb) __Pyx_ErrFetchInState(__pyx_tstate, type, value, tb) +static CYTHON_INLINE void __Pyx_ErrRestoreInState(PyThreadState *tstate, PyObject *type, PyObject *value, PyObject *tb); +static CYTHON_INLINE void __Pyx_ErrFetchInState(PyThreadState *tstate, PyObject **type, PyObject **value, PyObject **tb); +#if CYTHON_COMPILING_IN_CPYTHON +#define __Pyx_PyErr_SetNone(exc) (Py_INCREF(exc), __Pyx_ErrRestore((exc), NULL, NULL)) +#else +#define __Pyx_PyErr_SetNone(exc) PyErr_SetNone(exc) +#endif +#else +#define __Pyx_PyErr_Clear() PyErr_Clear() +#define __Pyx_PyErr_SetNone(exc) PyErr_SetNone(exc) +#define __Pyx_ErrRestoreWithState(type, value, tb) PyErr_Restore(type, value, tb) +#define __Pyx_ErrFetchWithState(type, value, tb) PyErr_Fetch(type, value, tb) +#define __Pyx_ErrRestoreInState(tstate, type, value, tb) PyErr_Restore(type, value, tb) +#define __Pyx_ErrFetchInState(tstate, type, value, tb) PyErr_Fetch(type, value, tb) +#define __Pyx_ErrRestore(type, value, tb) PyErr_Restore(type, value, tb) +#define __Pyx_ErrFetch(type, value, tb) PyErr_Fetch(type, value, tb) +#endif + +/* RaiseException.proto */ +static void __Pyx_Raise(PyObject *type, PyObject *value, PyObject *tb, PyObject *cause); + +/* PyCFunctionFastCall.proto */ +#if CYTHON_FAST_PYCCALL +static CYTHON_INLINE PyObject *__Pyx_PyCFunction_FastCall(PyObject *func, PyObject **args, Py_ssize_t nargs); +#else +#define __Pyx_PyCFunction_FastCall(func, args, nargs) (assert(0), NULL) +#endif + +/* PyFunctionFastCall.proto */ +#if CYTHON_FAST_PYCALL +#define __Pyx_PyFunction_FastCall(func, args, nargs)\ + __Pyx_PyFunction_FastCallDict((func), (args), (nargs), NULL) +#if 1 || PY_VERSION_HEX < 0x030600B1 +static PyObject *__Pyx_PyFunction_FastCallDict(PyObject *func, PyObject **args, Py_ssize_t nargs, PyObject *kwargs); +#else +#define __Pyx_PyFunction_FastCallDict(func, args, nargs, kwargs) _PyFunction_FastCallDict(func, args, nargs, kwargs) +#endif +#define __Pyx_BUILD_ASSERT_EXPR(cond)\ + (sizeof(char [1 - 2*!(cond)]) - 1) +#ifndef Py_MEMBER_SIZE +#define Py_MEMBER_SIZE(type, member) sizeof(((type *)0)->member) +#endif + static size_t __pyx_pyframe_localsplus_offset = 0; + #include "frameobject.h" + #define __Pxy_PyFrame_Initialize_Offsets()\ + ((void)__Pyx_BUILD_ASSERT_EXPR(sizeof(PyFrameObject) == offsetof(PyFrameObject, f_localsplus) + Py_MEMBER_SIZE(PyFrameObject, f_localsplus)),\ + (void)(__pyx_pyframe_localsplus_offset = ((size_t)PyFrame_Type.tp_basicsize) - Py_MEMBER_SIZE(PyFrameObject, f_localsplus))) + #define __Pyx_PyFrame_GetLocalsplus(frame)\ + (assert(__pyx_pyframe_localsplus_offset), (PyObject **)(((char *)(frame)) + __pyx_pyframe_localsplus_offset)) +#endif + +/* PyObjectCallMethO.proto */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallMethO(PyObject *func, PyObject *arg); +#endif + +/* PyObjectCallOneArg.proto */ +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallOneArg(PyObject *func, PyObject *arg); + +/* DictGetItem.proto */ +#if PY_MAJOR_VERSION >= 3 && !CYTHON_COMPILING_IN_PYPY +static PyObject *__Pyx_PyDict_GetItem(PyObject *d, PyObject* key); +#define __Pyx_PyObject_Dict_GetItem(obj, name)\ + (likely(PyDict_CheckExact(obj)) ?\ + __Pyx_PyDict_GetItem(obj, name) : PyObject_GetItem(obj, name)) +#else +#define __Pyx_PyDict_GetItem(d, key) PyObject_GetItem(d, key) +#define __Pyx_PyObject_Dict_GetItem(obj, name) PyObject_GetItem(obj, name) +#endif + +/* RaiseNoneIterError.proto */ +static CYTHON_INLINE void __Pyx_RaiseNoneNotIterableError(void); + +/* ExtTypeTest.proto */ +static CYTHON_INLINE int __Pyx_TypeTest(PyObject *obj, PyTypeObject *type); + +/* GetTopmostException.proto */ +#if CYTHON_USE_EXC_INFO_STACK +static _PyErr_StackItem * __Pyx_PyErr_GetTopmostException(PyThreadState *tstate); +#endif + +/* SaveResetException.proto */ +#if CYTHON_FAST_THREAD_STATE +#define __Pyx_ExceptionSave(type, value, tb) __Pyx__ExceptionSave(__pyx_tstate, type, value, tb) +static CYTHON_INLINE void __Pyx__ExceptionSave(PyThreadState *tstate, PyObject **type, PyObject **value, PyObject **tb); +#define __Pyx_ExceptionReset(type, value, tb) __Pyx__ExceptionReset(__pyx_tstate, type, value, tb) +static CYTHON_INLINE void __Pyx__ExceptionReset(PyThreadState *tstate, PyObject *type, PyObject *value, PyObject *tb); +#else +#define __Pyx_ExceptionSave(type, value, tb) PyErr_GetExcInfo(type, value, tb) +#define __Pyx_ExceptionReset(type, value, tb) PyErr_SetExcInfo(type, value, tb) +#endif + +/* PyErrExceptionMatches.proto */ +#if CYTHON_FAST_THREAD_STATE +#define __Pyx_PyErr_ExceptionMatches(err) __Pyx_PyErr_ExceptionMatchesInState(__pyx_tstate, err) +static CYTHON_INLINE int __Pyx_PyErr_ExceptionMatchesInState(PyThreadState* tstate, PyObject* err); +#else +#define __Pyx_PyErr_ExceptionMatches(err) PyErr_ExceptionMatches(err) +#endif + +/* GetException.proto */ +#if CYTHON_FAST_THREAD_STATE +#define __Pyx_GetException(type, value, tb) __Pyx__GetException(__pyx_tstate, type, value, tb) +static int __Pyx__GetException(PyThreadState *tstate, PyObject **type, PyObject **value, PyObject **tb); +#else +static int __Pyx_GetException(PyObject **type, PyObject **value, PyObject **tb); +#endif + +/* TypeImport.proto */ +#ifndef __PYX_HAVE_RT_ImportType_proto +#define __PYX_HAVE_RT_ImportType_proto +enum __Pyx_ImportType_CheckSize { + __Pyx_ImportType_CheckSize_Error = 0, + __Pyx_ImportType_CheckSize_Warn = 1, + __Pyx_ImportType_CheckSize_Ignore = 2 +}; +static PyTypeObject *__Pyx_ImportType(PyObject* module, const char *module_name, const char *class_name, size_t size, enum __Pyx_ImportType_CheckSize check_size); +#endif + +/* Import.proto */ +static PyObject *__Pyx_Import(PyObject *name, PyObject *from_list, int level); + +/* PyDictVersioning.proto */ +#if CYTHON_USE_DICT_VERSIONS && CYTHON_USE_TYPE_SLOTS +#define __PYX_DICT_VERSION_INIT ((PY_UINT64_T) -1) +#define __PYX_GET_DICT_VERSION(dict) (((PyDictObject*)(dict))->ma_version_tag) +#define __PYX_UPDATE_DICT_CACHE(dict, value, cache_var, version_var)\ + (version_var) = __PYX_GET_DICT_VERSION(dict);\ + (cache_var) = (value); +#define __PYX_PY_DICT_LOOKUP_IF_MODIFIED(VAR, DICT, LOOKUP) {\ + static PY_UINT64_T __pyx_dict_version = 0;\ + static PyObject *__pyx_dict_cached_value = NULL;\ + if (likely(__PYX_GET_DICT_VERSION(DICT) == __pyx_dict_version)) {\ + (VAR) = __pyx_dict_cached_value;\ + } else {\ + (VAR) = __pyx_dict_cached_value = (LOOKUP);\ + __pyx_dict_version = __PYX_GET_DICT_VERSION(DICT);\ + }\ +} +static CYTHON_INLINE PY_UINT64_T __Pyx_get_tp_dict_version(PyObject *obj); +static CYTHON_INLINE PY_UINT64_T __Pyx_get_object_dict_version(PyObject *obj); +static CYTHON_INLINE int __Pyx_object_dict_version_matches(PyObject* obj, PY_UINT64_T tp_dict_version, PY_UINT64_T obj_dict_version); +#else +#define __PYX_GET_DICT_VERSION(dict) (0) +#define __PYX_UPDATE_DICT_CACHE(dict, value, cache_var, version_var) +#define __PYX_PY_DICT_LOOKUP_IF_MODIFIED(VAR, DICT, LOOKUP) (VAR) = (LOOKUP); +#endif + +/* GetModuleGlobalName.proto */ +#if CYTHON_USE_DICT_VERSIONS +#define __Pyx_GetModuleGlobalName(var, name) {\ + static PY_UINT64_T __pyx_dict_version = 0;\ + static PyObject *__pyx_dict_cached_value = NULL;\ + (var) = (likely(__pyx_dict_version == __PYX_GET_DICT_VERSION(__pyx_d))) ?\ + (likely(__pyx_dict_cached_value) ? __Pyx_NewRef(__pyx_dict_cached_value) : __Pyx_GetBuiltinName(name)) :\ + __Pyx__GetModuleGlobalName(name, &__pyx_dict_version, &__pyx_dict_cached_value);\ +} +#define __Pyx_GetModuleGlobalNameUncached(var, name) {\ + PY_UINT64_T __pyx_dict_version;\ + PyObject *__pyx_dict_cached_value;\ + (var) = __Pyx__GetModuleGlobalName(name, &__pyx_dict_version, &__pyx_dict_cached_value);\ +} +static PyObject *__Pyx__GetModuleGlobalName(PyObject *name, PY_UINT64_T *dict_version, PyObject **dict_cached_value); +#else +#define __Pyx_GetModuleGlobalName(var, name) (var) = __Pyx__GetModuleGlobalName(name) +#define __Pyx_GetModuleGlobalNameUncached(var, name) (var) = __Pyx__GetModuleGlobalName(name) +static CYTHON_INLINE PyObject *__Pyx__GetModuleGlobalName(PyObject *name); +#endif + +/* PyObjectCallNoArg.proto */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallNoArg(PyObject *func); +#else +#define __Pyx_PyObject_CallNoArg(func) __Pyx_PyObject_Call(func, __pyx_empty_tuple, NULL) +#endif + +/* CLineInTraceback.proto */ +#ifdef CYTHON_CLINE_IN_TRACEBACK +#define __Pyx_CLineForTraceback(tstate, c_line) (((CYTHON_CLINE_IN_TRACEBACK)) ? c_line : 0) +#else +static int __Pyx_CLineForTraceback(PyThreadState *tstate, int c_line); +#endif + +/* CodeObjectCache.proto */ +typedef struct { + PyCodeObject* code_object; + int code_line; +} __Pyx_CodeObjectCacheEntry; +struct __Pyx_CodeObjectCache { + int count; + int max_count; + __Pyx_CodeObjectCacheEntry* entries; +}; +static struct __Pyx_CodeObjectCache __pyx_code_cache = {0,0,NULL}; +static int __pyx_bisect_code_objects(__Pyx_CodeObjectCacheEntry* entries, int count, int code_line); +static PyCodeObject *__pyx_find_code_object(int code_line); +static void __pyx_insert_code_object(int code_line, PyCodeObject* code_object); + +/* AddTraceback.proto */ +static void __Pyx_AddTraceback(const char *funcname, int c_line, + int py_line, const char *filename); + +/* CIntToPy.proto */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_npy_uint32(npy_uint32 value); + +/* RealImag.proto */ +#if CYTHON_CCOMPLEX + #ifdef __cplusplus + #define __Pyx_CREAL(z) ((z).real()) + #define __Pyx_CIMAG(z) ((z).imag()) + #else + #define __Pyx_CREAL(z) (__real__(z)) + #define __Pyx_CIMAG(z) (__imag__(z)) + #endif +#else + #define __Pyx_CREAL(z) ((z).real) + #define __Pyx_CIMAG(z) ((z).imag) +#endif +#if defined(__cplusplus) && CYTHON_CCOMPLEX\ + && (defined(_WIN32) || defined(__clang__) || (defined(__GNUC__) && (__GNUC__ >= 5 || __GNUC__ == 4 && __GNUC_MINOR__ >= 4 )) || __cplusplus >= 201103) + #define __Pyx_SET_CREAL(z,x) ((z).real(x)) + #define __Pyx_SET_CIMAG(z,y) ((z).imag(y)) +#else + #define __Pyx_SET_CREAL(z,x) __Pyx_CREAL(z) = (x) + #define __Pyx_SET_CIMAG(z,y) __Pyx_CIMAG(z) = (y) +#endif + +/* Arithmetic.proto */ +#if CYTHON_CCOMPLEX + #define __Pyx_c_eq_float(a, b) ((a)==(b)) + #define __Pyx_c_sum_float(a, b) ((a)+(b)) + #define __Pyx_c_diff_float(a, b) ((a)-(b)) + #define __Pyx_c_prod_float(a, b) ((a)*(b)) + #define __Pyx_c_quot_float(a, b) ((a)/(b)) + #define __Pyx_c_neg_float(a) (-(a)) + #ifdef __cplusplus + #define __Pyx_c_is_zero_float(z) ((z)==(float)0) + #define __Pyx_c_conj_float(z) (::std::conj(z)) + #if 1 + #define __Pyx_c_abs_float(z) (::std::abs(z)) + #define __Pyx_c_pow_float(a, b) (::std::pow(a, b)) + #endif + #else + #define __Pyx_c_is_zero_float(z) ((z)==0) + #define __Pyx_c_conj_float(z) (conjf(z)) + #if 1 + #define __Pyx_c_abs_float(z) (cabsf(z)) + #define __Pyx_c_pow_float(a, b) (cpowf(a, b)) + #endif + #endif +#else + static CYTHON_INLINE int __Pyx_c_eq_float(__pyx_t_float_complex, __pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_sum_float(__pyx_t_float_complex, __pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_diff_float(__pyx_t_float_complex, __pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_prod_float(__pyx_t_float_complex, __pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_quot_float(__pyx_t_float_complex, __pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_neg_float(__pyx_t_float_complex); + static CYTHON_INLINE int __Pyx_c_is_zero_float(__pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_conj_float(__pyx_t_float_complex); + #if 1 + static CYTHON_INLINE float __Pyx_c_abs_float(__pyx_t_float_complex); + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_pow_float(__pyx_t_float_complex, __pyx_t_float_complex); + #endif +#endif + +/* Arithmetic.proto */ +#if CYTHON_CCOMPLEX + #define __Pyx_c_eq_double(a, b) ((a)==(b)) + #define __Pyx_c_sum_double(a, b) ((a)+(b)) + #define __Pyx_c_diff_double(a, b) ((a)-(b)) + #define __Pyx_c_prod_double(a, b) ((a)*(b)) + #define __Pyx_c_quot_double(a, b) ((a)/(b)) + #define __Pyx_c_neg_double(a) (-(a)) + #ifdef __cplusplus + #define __Pyx_c_is_zero_double(z) ((z)==(double)0) + #define __Pyx_c_conj_double(z) (::std::conj(z)) + #if 1 + #define __Pyx_c_abs_double(z) (::std::abs(z)) + #define __Pyx_c_pow_double(a, b) (::std::pow(a, b)) + #endif + #else + #define __Pyx_c_is_zero_double(z) ((z)==0) + #define __Pyx_c_conj_double(z) (conj(z)) + #if 1 + #define __Pyx_c_abs_double(z) (cabs(z)) + #define __Pyx_c_pow_double(a, b) (cpow(a, b)) + #endif + #endif +#else + static CYTHON_INLINE int __Pyx_c_eq_double(__pyx_t_double_complex, __pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_sum_double(__pyx_t_double_complex, __pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_diff_double(__pyx_t_double_complex, __pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_prod_double(__pyx_t_double_complex, __pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_quot_double(__pyx_t_double_complex, __pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_neg_double(__pyx_t_double_complex); + static CYTHON_INLINE int __Pyx_c_is_zero_double(__pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_conj_double(__pyx_t_double_complex); + #if 1 + static CYTHON_INLINE double __Pyx_c_abs_double(__pyx_t_double_complex); + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_pow_double(__pyx_t_double_complex, __pyx_t_double_complex); + #endif +#endif + +/* CIntToPy.proto */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_int(int value); + +/* CIntToPy.proto */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_enum__NPY_TYPES(enum NPY_TYPES value); + +/* CIntFromPy.proto */ +static CYTHON_INLINE npy_uint32 __Pyx_PyInt_As_npy_uint32(PyObject *); + +/* CIntFromPy.proto */ +static CYTHON_INLINE int __Pyx_PyInt_As_int(PyObject *); + +/* CIntToPy.proto */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_long(long value); + +/* CIntFromPy.proto */ +static CYTHON_INLINE long __Pyx_PyInt_As_long(PyObject *); + +/* FastTypeChecks.proto */ +#if CYTHON_COMPILING_IN_CPYTHON +#define __Pyx_TypeCheck(obj, type) __Pyx_IsSubtype(Py_TYPE(obj), (PyTypeObject *)type) +static CYTHON_INLINE int __Pyx_IsSubtype(PyTypeObject *a, PyTypeObject *b); +static CYTHON_INLINE int __Pyx_PyErr_GivenExceptionMatches(PyObject *err, PyObject *type); +static CYTHON_INLINE int __Pyx_PyErr_GivenExceptionMatches2(PyObject *err, PyObject *type1, PyObject *type2); +#else +#define __Pyx_TypeCheck(obj, type) PyObject_TypeCheck(obj, (PyTypeObject *)type) +#define __Pyx_PyErr_GivenExceptionMatches(err, type) PyErr_GivenExceptionMatches(err, type) +#define __Pyx_PyErr_GivenExceptionMatches2(err, type1, type2) (PyErr_GivenExceptionMatches(err, type1) || PyErr_GivenExceptionMatches(err, type2)) +#endif +#define __Pyx_PyException_Check(obj) __Pyx_TypeCheck(obj, PyExc_Exception) + +/* CheckBinaryVersion.proto */ +static int __Pyx_check_binary_version(void); + +/* VoidPtrImport.proto */ +static int __Pyx_ImportVoidPtr(PyObject *module, const char *name, void **p, const char *sig); + +/* FunctionImport.proto */ +static int __Pyx_ImportFunction(PyObject *module, const char *funcname, void (**f)(void), const char *sig); + +/* InitStrings.proto */ +static int __Pyx_InitStrings(__Pyx_StringTabEntry *t); + + +/* Module declarations from 'cython' */ + +/* Module declarations from 'cpython.buffer' */ + +/* Module declarations from 'libc.string' */ + +/* Module declarations from 'libc.stdio' */ + +/* Module declarations from '__builtin__' */ + +/* Module declarations from 'cpython.type' */ +static PyTypeObject *__pyx_ptype_7cpython_4type_type = 0; + +/* Module declarations from 'cpython' */ + +/* Module declarations from 'cpython.object' */ + +/* Module declarations from 'cpython.ref' */ + +/* Module declarations from 'cpython.mem' */ + +/* Module declarations from 'numpy' */ + +/* Module declarations from 'numpy' */ +static PyTypeObject *__pyx_ptype_5numpy_dtype = 0; +static PyTypeObject *__pyx_ptype_5numpy_flatiter = 0; +static PyTypeObject *__pyx_ptype_5numpy_broadcast = 0; +static PyTypeObject *__pyx_ptype_5numpy_ndarray = 0; +static PyTypeObject *__pyx_ptype_5numpy_ufunc = 0; +static CYTHON_INLINE char *__pyx_f_5numpy__util_dtypestring(PyArray_Descr *, char *, char *, int *); /*proto*/ + +/* Module declarations from 'average_inner' */ +static __pyx_t_13average_inner_saxpy_ptr *__pyx_vp_13average_inner_saxpy = 0; +#define __pyx_v_13average_inner_saxpy (*__pyx_vp_13average_inner_saxpy) +static __pyx_t_13average_inner_sscal_ptr *__pyx_vp_13average_inner_sscal = 0; +#define __pyx_v_13average_inner_sscal (*__pyx_vp_13average_inner_sscal) +static int *__pyx_vp_13average_inner_ONE = 0; +#define __pyx_v_13average_inner_ONE (*__pyx_vp_13average_inner_ONE) +static int *__pyx_vp_13average_inner_ZERO = 0; +#define __pyx_v_13average_inner_ZERO (*__pyx_vp_13average_inner_ZERO) +static __pyx_t_13average_inner_REAL_t *__pyx_vp_13average_inner_ONEF = 0; +#define __pyx_v_13average_inner_ONEF (*__pyx_vp_13average_inner_ONEF) +static __pyx_t_13average_inner_REAL_t *__pyx_vp_13average_inner_ZEROF = 0; +#define __pyx_v_13average_inner_ZEROF (*__pyx_vp_13average_inner_ZEROF) +static PyObject *(*__pyx_f_13average_inner_init_base_s2v_config)(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *, PyObject *); /*proto*/ +static PyObject *(*__pyx_f_13average_inner_init_ft_s2v_config)(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *, PyObject *); /*proto*/ +static PyObject *(*__pyx_f_13average_inner_populate_base_s2v_config)(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *); /*proto*/ +static PyObject *(*__pyx_f_13average_inner_populate_ft_s2v_config)(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *); /*proto*/ + +/* Module declarations from 'pooling_inner' */ +static void __pyx_f_13pooling_inner_sl_max_pool(int const *, float *, float const *); /*proto*/ +static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ +#define __Pyx_MODULE_NAME "pooling_inner" +extern int __pyx_module_is_main_pooling_inner; +int __pyx_module_is_main_pooling_inner = 0; + +/* Implementation of 'pooling_inner' */ +static PyObject *__pyx_builtin_range; +static PyObject *__pyx_builtin_ValueError; +static PyObject *__pyx_builtin_RuntimeError; +static PyObject *__pyx_builtin_ImportError; +static const char __pyx_k__8[] = "*"; +static const char __pyx_k_ft[] = "ft"; +static const char __pyx_k_np[] = "np"; +static const char __pyx_k_wv[] = "wv"; +static const char __pyx_k_w2v[] = "w2v"; +static const char __pyx_k_init[] = "init"; +static const char __pyx_k_main[] = "__main__"; +static const char __pyx_k_name[] = "__name__"; +static const char __pyx_k_test[] = "__test__"; +static const char __pyx_k_fblas[] = "fblas"; +static const char __pyx_k_is_ft[] = "is_ft"; +static const char __pyx_k_model[] = "model"; +static const char __pyx_k_numpy[] = "numpy"; +static const char __pyx_k_range[] = "range"; +static const char __pyx_k_vocab[] = "vocab"; +static const char __pyx_k_import[] = "__import__"; +static const char __pyx_k_memory[] = "memory"; +static const char __pyx_k_target[] = "target"; +static const char __pyx_k_eff_words[] = "eff_words"; +static const char __pyx_k_ValueError[] = "ValueError"; +static const char __pyx_k_ImportError[] = "ImportError"; +static const char __pyx_k_FAST_VERSION[] = "FAST_VERSION"; +static const char __pyx_k_RuntimeError[] = "RuntimeError"; +static const char __pyx_k_eff_sentences[] = "eff_sentences"; +static const char __pyx_k_pooling_inner[] = "pooling_inner"; +static const char __pyx_k_train_pooling_cy[] = "train_pooling_cy"; +static const char __pyx_k_indexed_sentences[] = "indexed_sentences"; +static const char __pyx_k_pooling_inner_pyx[] = "pooling_inner.pyx"; +static const char __pyx_k_scipy_linalg_blas[] = "scipy.linalg.blas"; +static const char __pyx_k_MAX_WORDS_IN_BATCH[] = "MAX_WORDS_IN_BATCH"; +static const char __pyx_k_cline_in_traceback[] = "cline_in_traceback"; +static const char __pyx_k_MAX_NGRAMS_IN_BATCH[] = "MAX_NGRAMS_IN_BATCH"; +static const char __pyx_k_ndarray_is_not_C_contiguous[] = "ndarray is not C contiguous"; +static const char __pyx_k_numpy_core_multiarray_failed_to[] = "numpy.core.multiarray failed to import"; +static const char __pyx_k_unknown_dtype_code_in_numpy_pxd[] = "unknown dtype code in numpy.pxd (%d)"; +static const char __pyx_k_Format_string_allocated_too_shor[] = "Format string allocated too short, see comment in numpy.pxd"; +static const char __pyx_k_Non_native_byte_order_not_suppor[] = "Non-native byte order not supported"; +static const char __pyx_k_Optimized_cython_functions_for_c[] = "Optimized cython functions for computing sentence embeddings"; +static const char __pyx_k_ndarray_is_not_Fortran_contiguou[] = "ndarray is not Fortran contiguous"; +static const char __pyx_k_numpy_core_umath_failed_to_impor[] = "numpy.core.umath failed to import"; +static const char __pyx_k_Format_string_allocated_too_shor_2[] = "Format string allocated too short."; +static PyObject *__pyx_n_s_FAST_VERSION; +static PyObject *__pyx_kp_u_Format_string_allocated_too_shor; +static PyObject *__pyx_kp_u_Format_string_allocated_too_shor_2; +static PyObject *__pyx_n_s_ImportError; +static PyObject *__pyx_n_s_MAX_NGRAMS_IN_BATCH; +static PyObject *__pyx_n_s_MAX_WORDS_IN_BATCH; +static PyObject *__pyx_kp_u_Non_native_byte_order_not_suppor; +static PyObject *__pyx_n_s_RuntimeError; +static PyObject *__pyx_n_s_ValueError; +static PyObject *__pyx_n_s__8; +static PyObject *__pyx_n_s_cline_in_traceback; +static PyObject *__pyx_n_s_eff_sentences; +static PyObject *__pyx_n_s_eff_words; +static PyObject *__pyx_n_s_fblas; +static PyObject *__pyx_n_s_ft; +static PyObject *__pyx_n_s_import; +static PyObject *__pyx_n_s_indexed_sentences; +static PyObject *__pyx_n_s_init; +static PyObject *__pyx_n_s_is_ft; +static PyObject *__pyx_n_s_main; +static PyObject *__pyx_n_s_memory; +static PyObject *__pyx_n_s_model; +static PyObject *__pyx_n_s_name; +static PyObject *__pyx_kp_u_ndarray_is_not_C_contiguous; +static PyObject *__pyx_kp_u_ndarray_is_not_Fortran_contiguou; +static PyObject *__pyx_n_s_np; +static PyObject *__pyx_n_s_numpy; +static PyObject *__pyx_kp_s_numpy_core_multiarray_failed_to; +static PyObject *__pyx_kp_s_numpy_core_umath_failed_to_impor; +static PyObject *__pyx_n_s_pooling_inner; +static PyObject *__pyx_kp_s_pooling_inner_pyx; +static PyObject *__pyx_n_s_range; +static PyObject *__pyx_n_s_scipy_linalg_blas; +static PyObject *__pyx_n_s_target; +static PyObject *__pyx_n_s_test; +static PyObject *__pyx_n_s_train_pooling_cy; +static PyObject *__pyx_kp_u_unknown_dtype_code_in_numpy_pxd; +static PyObject *__pyx_n_s_vocab; +static PyObject *__pyx_n_s_w2v; +static PyObject *__pyx_n_s_wv; +static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObject *__pyx_self, PyObject *__pyx_v_model, PyObject *__pyx_v_indexed_sentences, PyObject *__pyx_v_target, PyObject *__pyx_v_memory); /* proto */ +static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_self); /* proto */ +static int __pyx_pf_5numpy_7ndarray___getbuffer__(PyArrayObject *__pyx_v_self, Py_buffer *__pyx_v_info, int __pyx_v_flags); /* proto */ +static void __pyx_pf_5numpy_7ndarray_2__releasebuffer__(PyArrayObject *__pyx_v_self, Py_buffer *__pyx_v_info); /* proto */ +static PyObject *__pyx_int_1; +static PyObject *__pyx_int_40; +static PyObject *__pyx_int_10000; +static PyObject *__pyx_tuple_; +static PyObject *__pyx_tuple__2; +static PyObject *__pyx_tuple__3; +static PyObject *__pyx_tuple__4; +static PyObject *__pyx_tuple__5; +static PyObject *__pyx_tuple__6; +static PyObject *__pyx_tuple__7; +static PyObject *__pyx_tuple__9; +static PyObject *__pyx_codeobj__10; +static PyObject *__pyx_codeobj__11; +/* Late includes */ + +/* "pooling_inner.pyx":42 + * DEF MAX_NGRAMS = 40 + * + * cdef void sl_max_pool(const int *N, float *X, const float *Y) nogil: # <<<<<<<<<<<<<< + * """ Performs single left max pooling op + * + */ + +static void __pyx_f_13pooling_inner_sl_max_pool(int const *__pyx_v_N, float *__pyx_v_X, float const *__pyx_v_Y) { + int __pyx_v_i; + int __pyx_t_1; + int __pyx_t_2; + + /* "pooling_inner.pyx":56 + * """ + * cdef int i + * for i from 0 <= i < N[0] by 1: # <<<<<<<<<<<<<< + * if X[i] < Y[i]: + * X[i] = Y[i] + */ + __pyx_t_1 = (__pyx_v_N[0]); + for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_1; __pyx_v_i+=1) { + + /* "pooling_inner.pyx":57 + * cdef int i + * for i from 0 <= i < N[0] by 1: + * if X[i] < Y[i]: # <<<<<<<<<<<<<< + * X[i] = Y[i] + * + */ + __pyx_t_2 = (((__pyx_v_X[__pyx_v_i]) < (__pyx_v_Y[__pyx_v_i])) != 0); + if (__pyx_t_2) { + + /* "pooling_inner.pyx":58 + * for i from 0 <= i < N[0] by 1: + * if X[i] < Y[i]: + * X[i] = Y[i] # <<<<<<<<<<<<<< + * + * cdef void compute_base_sentence_pooling(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: + */ + (__pyx_v_X[__pyx_v_i]) = (__pyx_v_Y[__pyx_v_i]); + + /* "pooling_inner.pyx":57 + * cdef int i + * for i from 0 <= i < N[0] by 1: + * if X[i] < Y[i]: # <<<<<<<<<<<<<< + * X[i] = Y[i] + * + */ + } + } + + /* "pooling_inner.pyx":42 + * DEF MAX_NGRAMS = 40 + * + * cdef void sl_max_pool(const int *N, float *X, const float *Y) nogil: # <<<<<<<<<<<<<< + * """ Performs single left max pooling op + * + */ + + /* function exit code */ +} + +/* "pooling_inner.pyx":60 + * X[i] = Y[i] + * + * cdef void compute_base_sentence_pooling(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< + * """Perform optimized sentence-level max pooling for BaseAny2Vec model. + * + */ + +static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { + int __pyx_v_size; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_end; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; + __pyx_t_13average_inner_uINT_t __pyx_v_i; + __pyx_t_13average_inner_uINT_t __pyx_v_word_row; + __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; + int __pyx_t_1; + __pyx_t_13average_inner_uINT_t __pyx_t_2; + __pyx_t_13average_inner_uINT_t __pyx_t_3; + __pyx_t_13average_inner_uINT_t __pyx_t_4; + __pyx_t_13average_inner_uINT_t __pyx_t_5; + __pyx_t_13average_inner_uINT_t __pyx_t_6; + __pyx_t_13average_inner_uINT_t __pyx_t_7; + + /* "pooling_inner.pyx":76 + * """ + * cdef: + * int size = c.size # <<<<<<<<<<<<<< + * + * uINT_t sent_idx, sent_start, sent_end, sent_row + */ + __pyx_t_1 = __pyx_v_c->size; + __pyx_v_size = __pyx_t_1; + + /* "pooling_inner.pyx":84 + * REAL_t sent_len, inv_count + * + * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + */ + __pyx_t_2 = __pyx_v_num_sentences; + __pyx_t_3 = __pyx_t_2; + for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) { + __pyx_v_sent_idx = __pyx_t_4; + + /* "pooling_inner.pyx":85 + * + * for sent_idx in range(num_sentences): + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + * + */ + (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); + + /* "pooling_inner.pyx":86 + * for sent_idx in range(num_sentences): + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< + * + * sent_start = c.sentence_boundary[sent_idx] + */ + (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); + + /* "pooling_inner.pyx":88 + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + * + * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< + * sent_end = c.sentence_boundary[sent_idx + 1] + * sent_len = ZEROF + */ + __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); + + /* "pooling_inner.pyx":89 + * + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< + * sent_len = ZEROF + * + */ + __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); + + /* "pooling_inner.pyx":90 + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] + * sent_len = ZEROF # <<<<<<<<<<<<<< + * + * for i in range(sent_start, sent_end): + */ + __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; + + /* "pooling_inner.pyx":92 + * sent_len = ZEROF + * + * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< + * sent_len += ONEF + * sent_row = c.sent_adresses[i] * size + */ + __pyx_t_5 = __pyx_v_sent_end; + __pyx_t_6 = __pyx_t_5; + for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { + __pyx_v_i = __pyx_t_7; + + /* "pooling_inner.pyx":93 + * + * for i in range(sent_start, sent_end): + * sent_len += ONEF # <<<<<<<<<<<<<< + * sent_row = c.sent_adresses[i] * size + * word_row = c.word_indices[i] * size + */ + __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); + + /* "pooling_inner.pyx":94 + * for i in range(sent_start, sent_end): + * sent_len += ONEF + * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< + * word_row = c.word_indices[i] * size + * + */ + __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); + + /* "pooling_inner.pyx":95 + * sent_len += ONEF + * sent_row = c.sent_adresses[i] * size + * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< + * + * sl_max_pool( + */ + __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); + + /* "pooling_inner.pyx":97 + * word_row = c.word_indices[i] * size + * + * sl_max_pool( # <<<<<<<<<<<<<< + * &size, + * &c.sentence_vectors[sent_row], + */ + __pyx_f_13pooling_inner_sl_max_pool((&__pyx_v_size), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row]))); + } + } + + /* "pooling_inner.pyx":60 + * X[i] = Y[i] + * + * cdef void compute_base_sentence_pooling(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< + * """Perform optimized sentence-level max pooling for BaseAny2Vec model. + * + */ + + /* function exit code */ +} + +/* "pooling_inner.pyx":160 + * # saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + * + * def train_pooling_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< + * """Training on a sequence of sentences and update the target ndarray. + * + */ + +/* Python wrapper */ +static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/ +static char __pyx_doc_13pooling_inner_train_pooling_cy[] = "train_pooling_cy(model, indexed_sentences, target, memory)\nTraining on a sequence of sentences and update the target ndarray.\n\n Called internally from :meth:`~fse.models.pooling.MaxPooling._do_train_job`.\n\n Parameters\n ----------\n model : :class:`~fse.models.base_s2v.BaseSentence2VecModel`\n The BaseSentence2VecModel model instance.\n indexed_sentences : iterable of tuple\n The sentences used to train the model.\n target : ndarray\n The target ndarray. We use the index from indexed_sentences\n to write into the corresponding row of target.\n memory : ndarray\n Private memory for each working thread.\n\n Returns\n -------\n int, int\n Number of effective sentences (non-zero) and effective words in the vocabulary used \n during training the sentence embedding.\n "; +static PyMethodDef __pyx_mdef_13pooling_inner_1train_pooling_cy = {"train_pooling_cy", (PyCFunction)(void*)(PyCFunctionWithKeywords)__pyx_pw_13pooling_inner_1train_pooling_cy, METH_VARARGS|METH_KEYWORDS, __pyx_doc_13pooling_inner_train_pooling_cy}; +static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self, PyObject *__pyx_args, PyObject *__pyx_kwds) { + PyObject *__pyx_v_model = 0; + PyObject *__pyx_v_indexed_sentences = 0; + PyObject *__pyx_v_target = 0; + PyObject *__pyx_v_memory = 0; + PyObject *__pyx_r = 0; + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("train_pooling_cy (wrapper)", 0); + { + static PyObject **__pyx_pyargnames[] = {&__pyx_n_s_model,&__pyx_n_s_indexed_sentences,&__pyx_n_s_target,&__pyx_n_s_memory,0}; + PyObject* values[4] = {0,0,0,0}; + if (unlikely(__pyx_kwds)) { + Py_ssize_t kw_args; + const Py_ssize_t pos_args = PyTuple_GET_SIZE(__pyx_args); + switch (pos_args) { + case 4: values[3] = PyTuple_GET_ITEM(__pyx_args, 3); + CYTHON_FALLTHROUGH; + case 3: values[2] = PyTuple_GET_ITEM(__pyx_args, 2); + CYTHON_FALLTHROUGH; + case 2: values[1] = PyTuple_GET_ITEM(__pyx_args, 1); + CYTHON_FALLTHROUGH; + case 1: values[0] = PyTuple_GET_ITEM(__pyx_args, 0); + CYTHON_FALLTHROUGH; + case 0: break; + default: goto __pyx_L5_argtuple_error; + } + kw_args = PyDict_Size(__pyx_kwds); + switch (pos_args) { + case 0: + if (likely((values[0] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_model)) != 0)) kw_args--; + else goto __pyx_L5_argtuple_error; + CYTHON_FALLTHROUGH; + case 1: + if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; + else { + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 160, __pyx_L3_error) + } + CYTHON_FALLTHROUGH; + case 2: + if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; + else { + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 160, __pyx_L3_error) + } + CYTHON_FALLTHROUGH; + case 3: + if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; + else { + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 160, __pyx_L3_error) + } + } + if (unlikely(kw_args > 0)) { + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 160, __pyx_L3_error) + } + } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { + goto __pyx_L5_argtuple_error; + } else { + values[0] = PyTuple_GET_ITEM(__pyx_args, 0); + values[1] = PyTuple_GET_ITEM(__pyx_args, 1); + values[2] = PyTuple_GET_ITEM(__pyx_args, 2); + values[3] = PyTuple_GET_ITEM(__pyx_args, 3); + } + __pyx_v_model = values[0]; + __pyx_v_indexed_sentences = values[1]; + __pyx_v_target = values[2]; + __pyx_v_memory = values[3]; + } + goto __pyx_L4_argument_unpacking_done; + __pyx_L5_argtuple_error:; + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 160, __pyx_L3_error) + __pyx_L3_error:; + __Pyx_AddTraceback("pooling_inner.train_pooling_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); + __Pyx_RefNannyFinishContext(); + return NULL; + __pyx_L4_argument_unpacking_done:; + __pyx_r = __pyx_pf_13pooling_inner_train_pooling_cy(__pyx_self, __pyx_v_model, __pyx_v_indexed_sentences, __pyx_v_target, __pyx_v_memory); + + /* function exit code */ + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObject *__pyx_self, PyObject *__pyx_v_model, PyObject *__pyx_v_indexed_sentences, PyObject *__pyx_v_target, PyObject *__pyx_v_memory) { + __pyx_t_13average_inner_uINT_t __pyx_v_eff_sentences; + __pyx_t_13average_inner_uINT_t __pyx_v_eff_words; + struct __pyx_t_13average_inner_BaseSentenceVecsConfig __pyx_v_w2v; + struct __pyx_t_13average_inner_FTSentenceVecsConfig __pyx_v_ft; + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + int __pyx_t_2; + int __pyx_t_3; + PyObject *__pyx_t_4 = NULL; + PyObject *__pyx_t_5 = NULL; + PyObject *__pyx_t_6 = NULL; + PyObject *(*__pyx_t_7)(PyObject *); + __pyx_t_13average_inner_uINT_t __pyx_t_8; + __pyx_t_13average_inner_uINT_t __pyx_t_9; + __Pyx_RefNannySetupContext("train_pooling_cy", 0); + + /* "pooling_inner.pyx":184 + * """ + * + * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< + * cdef uINT_t eff_words = 0 + * cdef BaseSentenceVecsConfig w2v + */ + __pyx_v_eff_sentences = 0; + + /* "pooling_inner.pyx":185 + * + * cdef uINT_t eff_sentences = 0 + * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< + * cdef BaseSentenceVecsConfig w2v + * cdef FTSentenceVecsConfig ft + */ + __pyx_v_eff_words = 0; + + /* "pooling_inner.pyx":189 + * cdef FTSentenceVecsConfig ft + * + * if not model.is_ft: # <<<<<<<<<<<<<< + * init_base_s2v_config(&w2v, model, target, memory) + * + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 189, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 189, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_3 = ((!__pyx_t_2) != 0); + if (__pyx_t_3) { + + /* "pooling_inner.pyx":190 + * + * if not model.is_ft: + * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< + * + * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + */ + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 190, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "pooling_inner.pyx":192 + * init_base_s2v_config(&w2v, model, target, memory) + * + * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) # <<<<<<<<<<<<<< + * + * with nogil: + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 192, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 192, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 192, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { + PyObject* sequence = __pyx_t_1; + Py_ssize_t size = __Pyx_PySequence_SIZE(sequence); + if (unlikely(size != 2)) { + if (size > 2) __Pyx_RaiseTooManyValuesError(2); + else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); + __PYX_ERR(0, 192, __pyx_L1_error) + } + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + if (likely(PyTuple_CheckExact(sequence))) { + __pyx_t_4 = PyTuple_GET_ITEM(sequence, 0); + __pyx_t_5 = PyTuple_GET_ITEM(sequence, 1); + } else { + __pyx_t_4 = PyList_GET_ITEM(sequence, 0); + __pyx_t_5 = PyList_GET_ITEM(sequence, 1); + } + __Pyx_INCREF(__pyx_t_4); + __Pyx_INCREF(__pyx_t_5); + #else + __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 192, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 192, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_5); + #endif + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + } else { + Py_ssize_t index = -1; + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 192, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_6); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; + index = 0; __pyx_t_4 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_4)) goto __pyx_L4_unpacking_failed; + __Pyx_GOTREF(__pyx_t_4); + index = 1; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L4_unpacking_failed; + __Pyx_GOTREF(__pyx_t_5); + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 192, __pyx_L1_error) + __pyx_t_7 = NULL; + __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; + goto __pyx_L5_unpacking_done; + __pyx_L4_unpacking_failed:; + __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; + __pyx_t_7 = NULL; + if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); + __PYX_ERR(0, 192, __pyx_L1_error) + __pyx_L5_unpacking_done:; + } + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 192, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 192, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; + __pyx_v_eff_sentences = __pyx_t_8; + __pyx_v_eff_words = __pyx_t_9; + + /* "pooling_inner.pyx":194 + * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + * + * with nogil: # <<<<<<<<<<<<<< + * compute_base_sentence_pooling(&w2v, eff_sentences) + * else: + */ + { + #ifdef WITH_THREAD + PyThreadState *_save; + Py_UNBLOCK_THREADS + __Pyx_FastGIL_Remember(); + #endif + /*try:*/ { + + /* "pooling_inner.pyx":195 + * + * with nogil: + * compute_base_sentence_pooling(&w2v, eff_sentences) # <<<<<<<<<<<<<< + * else: + * init_ft_s2v_config(&ft, model, target, memory) + */ + __pyx_f_13pooling_inner_compute_base_sentence_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences); + } + + /* "pooling_inner.pyx":194 + * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + * + * with nogil: # <<<<<<<<<<<<<< + * compute_base_sentence_pooling(&w2v, eff_sentences) + * else: + */ + /*finally:*/ { + /*normal exit:*/{ + #ifdef WITH_THREAD + __Pyx_FastGIL_Forget(); + Py_BLOCK_THREADS + #endif + goto __pyx_L8; + } + __pyx_L8:; + } + } + + /* "pooling_inner.pyx":189 + * cdef FTSentenceVecsConfig ft + * + * if not model.is_ft: # <<<<<<<<<<<<<< + * init_base_s2v_config(&w2v, model, target, memory) + * + */ + goto __pyx_L3; + } + + /* "pooling_inner.pyx":197 + * compute_base_sentence_pooling(&w2v, eff_sentences) + * else: + * init_ft_s2v_config(&ft, model, target, memory) # <<<<<<<<<<<<<< + * + * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + */ + /*else*/ { + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 197, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "pooling_inner.pyx":199 + * init_ft_s2v_config(&ft, model, target, memory) + * + * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) # <<<<<<<<<<<<<< + * + * # with nogil: + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 199, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 199, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_5); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 199, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; + if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { + PyObject* sequence = __pyx_t_1; + Py_ssize_t size = __Pyx_PySequence_SIZE(sequence); + if (unlikely(size != 2)) { + if (size > 2) __Pyx_RaiseTooManyValuesError(2); + else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); + __PYX_ERR(0, 199, __pyx_L1_error) + } + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + if (likely(PyTuple_CheckExact(sequence))) { + __pyx_t_5 = PyTuple_GET_ITEM(sequence, 0); + __pyx_t_4 = PyTuple_GET_ITEM(sequence, 1); + } else { + __pyx_t_5 = PyList_GET_ITEM(sequence, 0); + __pyx_t_4 = PyList_GET_ITEM(sequence, 1); + } + __Pyx_INCREF(__pyx_t_5); + __Pyx_INCREF(__pyx_t_4); + #else + __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 199, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_5); + __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 199, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + #endif + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + } else { + Py_ssize_t index = -1; + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 199, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_6); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; + index = 0; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L9_unpacking_failed; + __Pyx_GOTREF(__pyx_t_5); + index = 1; __pyx_t_4 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_4)) goto __pyx_L9_unpacking_failed; + __Pyx_GOTREF(__pyx_t_4); + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 199, __pyx_L1_error) + __pyx_t_7 = NULL; + __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; + goto __pyx_L10_unpacking_done; + __pyx_L9_unpacking_failed:; + __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; + __pyx_t_7 = NULL; + if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); + __PYX_ERR(0, 199, __pyx_L1_error) + __pyx_L10_unpacking_done:; + } + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 199, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 199, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_v_eff_sentences = __pyx_t_9; + __pyx_v_eff_words = __pyx_t_8; + } + __pyx_L3:; + + /* "pooling_inner.pyx":204 + * # compute_ft_sentence_averages(&ft, eff_sentences) + * + * return eff_sentences, eff_words # <<<<<<<<<<<<<< + * + * def init(): + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 204, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 204, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 204, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_5); + __Pyx_GIVEREF(__pyx_t_1); + PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_1); + __Pyx_GIVEREF(__pyx_t_4); + PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_4); + __pyx_t_1 = 0; + __pyx_t_4 = 0; + __pyx_r = __pyx_t_5; + __pyx_t_5 = 0; + goto __pyx_L0; + + /* "pooling_inner.pyx":160 + * # saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + * + * def train_pooling_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< + * """Training on a sequence of sentences and update the target ndarray. + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_XDECREF(__pyx_t_4); + __Pyx_XDECREF(__pyx_t_5); + __Pyx_XDECREF(__pyx_t_6); + __Pyx_AddTraceback("pooling_inner.train_pooling_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = NULL; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "pooling_inner.pyx":206 + * return eff_sentences, eff_words + * + * def init(): # <<<<<<<<<<<<<< + * return 1 + * + */ + +/* Python wrapper */ +static PyObject *__pyx_pw_13pooling_inner_3init(PyObject *__pyx_self, CYTHON_UNUSED PyObject *unused); /*proto*/ +static char __pyx_doc_13pooling_inner_2init[] = "init()"; +static PyMethodDef __pyx_mdef_13pooling_inner_3init = {"init", (PyCFunction)__pyx_pw_13pooling_inner_3init, METH_NOARGS, __pyx_doc_13pooling_inner_2init}; +static PyObject *__pyx_pw_13pooling_inner_3init(PyObject *__pyx_self, CYTHON_UNUSED PyObject *unused) { + PyObject *__pyx_r = 0; + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("init (wrapper)", 0); + __pyx_r = __pyx_pf_13pooling_inner_2init(__pyx_self); + + /* function exit code */ + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_self) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("init", 0); + + /* "pooling_inner.pyx":207 + * + * def init(): + * return 1 # <<<<<<<<<<<<<< + * + * MAX_WORDS_IN_BATCH = MAX_WORDS + */ + __Pyx_XDECREF(__pyx_r); + __Pyx_INCREF(__pyx_int_1); + __pyx_r = __pyx_int_1; + goto __pyx_L0; + + /* "pooling_inner.pyx":206 + * return eff_sentences, eff_words + * + * def init(): # <<<<<<<<<<<<<< + * return 1 + * + */ + + /* function exit code */ + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":258 + * # experimental exception made for __getbuffer__ and __releasebuffer__ + * # -- the details of this may change. + * def __getbuffer__(ndarray self, Py_buffer* info, int flags): # <<<<<<<<<<<<<< + * # This implementation of getbuffer is geared towards Cython + * # requirements, and does not yet fulfill the PEP. + */ + +/* Python wrapper */ +static CYTHON_UNUSED int __pyx_pw_5numpy_7ndarray_1__getbuffer__(PyObject *__pyx_v_self, Py_buffer *__pyx_v_info, int __pyx_v_flags); /*proto*/ +static CYTHON_UNUSED int __pyx_pw_5numpy_7ndarray_1__getbuffer__(PyObject *__pyx_v_self, Py_buffer *__pyx_v_info, int __pyx_v_flags) { + int __pyx_r; + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__getbuffer__ (wrapper)", 0); + __pyx_r = __pyx_pf_5numpy_7ndarray___getbuffer__(((PyArrayObject *)__pyx_v_self), ((Py_buffer *)__pyx_v_info), ((int)__pyx_v_flags)); + + /* function exit code */ + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +static int __pyx_pf_5numpy_7ndarray___getbuffer__(PyArrayObject *__pyx_v_self, Py_buffer *__pyx_v_info, int __pyx_v_flags) { + int __pyx_v_i; + int __pyx_v_ndim; + int __pyx_v_endian_detector; + int __pyx_v_little_endian; + int __pyx_v_t; + char *__pyx_v_f; + PyArray_Descr *__pyx_v_descr = 0; + int __pyx_v_offset; + int __pyx_r; + __Pyx_RefNannyDeclarations + int __pyx_t_1; + int __pyx_t_2; + PyObject *__pyx_t_3 = NULL; + int __pyx_t_4; + int __pyx_t_5; + int __pyx_t_6; + PyArray_Descr *__pyx_t_7; + PyObject *__pyx_t_8 = NULL; + char *__pyx_t_9; + if (__pyx_v_info == NULL) { + PyErr_SetString(PyExc_BufferError, "PyObject_GetBuffer: view==NULL argument is obsolete"); + return -1; + } + __Pyx_RefNannySetupContext("__getbuffer__", 0); + __pyx_v_info->obj = Py_None; __Pyx_INCREF(Py_None); + __Pyx_GIVEREF(__pyx_v_info->obj); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":265 + * + * cdef int i, ndim + * cdef int endian_detector = 1 # <<<<<<<<<<<<<< + * cdef bint little_endian = ((&endian_detector)[0] != 0) + * + */ + __pyx_v_endian_detector = 1; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":266 + * cdef int i, ndim + * cdef int endian_detector = 1 + * cdef bint little_endian = ((&endian_detector)[0] != 0) # <<<<<<<<<<<<<< + * + * ndim = PyArray_NDIM(self) + */ + __pyx_v_little_endian = ((((char *)(&__pyx_v_endian_detector))[0]) != 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":268 + * cdef bint little_endian = ((&endian_detector)[0] != 0) + * + * ndim = PyArray_NDIM(self) # <<<<<<<<<<<<<< + * + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) + */ + __pyx_v_ndim = PyArray_NDIM(__pyx_v_self); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":270 + * ndim = PyArray_NDIM(self) + * + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) # <<<<<<<<<<<<<< + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_C_CONTIGUOUS)): + * raise ValueError(u"ndarray is not C contiguous") + */ + __pyx_t_2 = (((__pyx_v_flags & PyBUF_C_CONTIGUOUS) == PyBUF_C_CONTIGUOUS) != 0); + if (__pyx_t_2) { + } else { + __pyx_t_1 = __pyx_t_2; + goto __pyx_L4_bool_binop_done; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":271 + * + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_C_CONTIGUOUS)): # <<<<<<<<<<<<<< + * raise ValueError(u"ndarray is not C contiguous") + * + */ + __pyx_t_2 = ((!(PyArray_CHKFLAGS(__pyx_v_self, NPY_ARRAY_C_CONTIGUOUS) != 0)) != 0); + __pyx_t_1 = __pyx_t_2; + __pyx_L4_bool_binop_done:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":270 + * ndim = PyArray_NDIM(self) + * + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) # <<<<<<<<<<<<<< + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_C_CONTIGUOUS)): + * raise ValueError(u"ndarray is not C contiguous") + */ + if (unlikely(__pyx_t_1)) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":272 + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_C_CONTIGUOUS)): + * raise ValueError(u"ndarray is not C contiguous") # <<<<<<<<<<<<<< + * + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) + */ + __pyx_t_3 = __Pyx_PyObject_Call(__pyx_builtin_ValueError, __pyx_tuple_, NULL); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 272, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_Raise(__pyx_t_3, 0, 0, 0); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __PYX_ERR(1, 272, __pyx_L1_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":270 + * ndim = PyArray_NDIM(self) + * + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) # <<<<<<<<<<<<<< + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_C_CONTIGUOUS)): + * raise ValueError(u"ndarray is not C contiguous") + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":274 + * raise ValueError(u"ndarray is not C contiguous") + * + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) # <<<<<<<<<<<<<< + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_F_CONTIGUOUS)): + * raise ValueError(u"ndarray is not Fortran contiguous") + */ + __pyx_t_2 = (((__pyx_v_flags & PyBUF_F_CONTIGUOUS) == PyBUF_F_CONTIGUOUS) != 0); + if (__pyx_t_2) { + } else { + __pyx_t_1 = __pyx_t_2; + goto __pyx_L7_bool_binop_done; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":275 + * + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_F_CONTIGUOUS)): # <<<<<<<<<<<<<< + * raise ValueError(u"ndarray is not Fortran contiguous") + * + */ + __pyx_t_2 = ((!(PyArray_CHKFLAGS(__pyx_v_self, NPY_ARRAY_F_CONTIGUOUS) != 0)) != 0); + __pyx_t_1 = __pyx_t_2; + __pyx_L7_bool_binop_done:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":274 + * raise ValueError(u"ndarray is not C contiguous") + * + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) # <<<<<<<<<<<<<< + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_F_CONTIGUOUS)): + * raise ValueError(u"ndarray is not Fortran contiguous") + */ + if (unlikely(__pyx_t_1)) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":276 + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_F_CONTIGUOUS)): + * raise ValueError(u"ndarray is not Fortran contiguous") # <<<<<<<<<<<<<< + * + * info.buf = PyArray_DATA(self) + */ + __pyx_t_3 = __Pyx_PyObject_Call(__pyx_builtin_ValueError, __pyx_tuple__2, NULL); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 276, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_Raise(__pyx_t_3, 0, 0, 0); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __PYX_ERR(1, 276, __pyx_L1_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":274 + * raise ValueError(u"ndarray is not C contiguous") + * + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) # <<<<<<<<<<<<<< + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_F_CONTIGUOUS)): + * raise ValueError(u"ndarray is not Fortran contiguous") + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":278 + * raise ValueError(u"ndarray is not Fortran contiguous") + * + * info.buf = PyArray_DATA(self) # <<<<<<<<<<<<<< + * info.ndim = ndim + * if sizeof(npy_intp) != sizeof(Py_ssize_t): + */ + __pyx_v_info->buf = PyArray_DATA(__pyx_v_self); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":279 + * + * info.buf = PyArray_DATA(self) + * info.ndim = ndim # <<<<<<<<<<<<<< + * if sizeof(npy_intp) != sizeof(Py_ssize_t): + * # Allocate new buffer for strides and shape info. + */ + __pyx_v_info->ndim = __pyx_v_ndim; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":280 + * info.buf = PyArray_DATA(self) + * info.ndim = ndim + * if sizeof(npy_intp) != sizeof(Py_ssize_t): # <<<<<<<<<<<<<< + * # Allocate new buffer for strides and shape info. + * # This is allocated as one block, strides first. + */ + __pyx_t_1 = (((sizeof(npy_intp)) != (sizeof(Py_ssize_t))) != 0); + if (__pyx_t_1) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":283 + * # Allocate new buffer for strides and shape info. + * # This is allocated as one block, strides first. + * info.strides = PyObject_Malloc(sizeof(Py_ssize_t) * 2 * ndim) # <<<<<<<<<<<<<< + * info.shape = info.strides + ndim + * for i in range(ndim): + */ + __pyx_v_info->strides = ((Py_ssize_t *)PyObject_Malloc((((sizeof(Py_ssize_t)) * 2) * ((size_t)__pyx_v_ndim)))); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":284 + * # This is allocated as one block, strides first. + * info.strides = PyObject_Malloc(sizeof(Py_ssize_t) * 2 * ndim) + * info.shape = info.strides + ndim # <<<<<<<<<<<<<< + * for i in range(ndim): + * info.strides[i] = PyArray_STRIDES(self)[i] + */ + __pyx_v_info->shape = (__pyx_v_info->strides + __pyx_v_ndim); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":285 + * info.strides = PyObject_Malloc(sizeof(Py_ssize_t) * 2 * ndim) + * info.shape = info.strides + ndim + * for i in range(ndim): # <<<<<<<<<<<<<< + * info.strides[i] = PyArray_STRIDES(self)[i] + * info.shape[i] = PyArray_DIMS(self)[i] + */ + __pyx_t_4 = __pyx_v_ndim; + __pyx_t_5 = __pyx_t_4; + for (__pyx_t_6 = 0; __pyx_t_6 < __pyx_t_5; __pyx_t_6+=1) { + __pyx_v_i = __pyx_t_6; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":286 + * info.shape = info.strides + ndim + * for i in range(ndim): + * info.strides[i] = PyArray_STRIDES(self)[i] # <<<<<<<<<<<<<< + * info.shape[i] = PyArray_DIMS(self)[i] + * else: + */ + (__pyx_v_info->strides[__pyx_v_i]) = (PyArray_STRIDES(__pyx_v_self)[__pyx_v_i]); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":287 + * for i in range(ndim): + * info.strides[i] = PyArray_STRIDES(self)[i] + * info.shape[i] = PyArray_DIMS(self)[i] # <<<<<<<<<<<<<< + * else: + * info.strides = PyArray_STRIDES(self) + */ + (__pyx_v_info->shape[__pyx_v_i]) = (PyArray_DIMS(__pyx_v_self)[__pyx_v_i]); + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":280 + * info.buf = PyArray_DATA(self) + * info.ndim = ndim + * if sizeof(npy_intp) != sizeof(Py_ssize_t): # <<<<<<<<<<<<<< + * # Allocate new buffer for strides and shape info. + * # This is allocated as one block, strides first. + */ + goto __pyx_L9; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":289 + * info.shape[i] = PyArray_DIMS(self)[i] + * else: + * info.strides = PyArray_STRIDES(self) # <<<<<<<<<<<<<< + * info.shape = PyArray_DIMS(self) + * info.suboffsets = NULL + */ + /*else*/ { + __pyx_v_info->strides = ((Py_ssize_t *)PyArray_STRIDES(__pyx_v_self)); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":290 + * else: + * info.strides = PyArray_STRIDES(self) + * info.shape = PyArray_DIMS(self) # <<<<<<<<<<<<<< + * info.suboffsets = NULL + * info.itemsize = PyArray_ITEMSIZE(self) + */ + __pyx_v_info->shape = ((Py_ssize_t *)PyArray_DIMS(__pyx_v_self)); + } + __pyx_L9:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":291 + * info.strides = PyArray_STRIDES(self) + * info.shape = PyArray_DIMS(self) + * info.suboffsets = NULL # <<<<<<<<<<<<<< + * info.itemsize = PyArray_ITEMSIZE(self) + * info.readonly = not PyArray_ISWRITEABLE(self) + */ + __pyx_v_info->suboffsets = NULL; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":292 + * info.shape = PyArray_DIMS(self) + * info.suboffsets = NULL + * info.itemsize = PyArray_ITEMSIZE(self) # <<<<<<<<<<<<<< + * info.readonly = not PyArray_ISWRITEABLE(self) + * + */ + __pyx_v_info->itemsize = PyArray_ITEMSIZE(__pyx_v_self); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":293 + * info.suboffsets = NULL + * info.itemsize = PyArray_ITEMSIZE(self) + * info.readonly = not PyArray_ISWRITEABLE(self) # <<<<<<<<<<<<<< + * + * cdef int t + */ + __pyx_v_info->readonly = (!(PyArray_ISWRITEABLE(__pyx_v_self) != 0)); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":296 + * + * cdef int t + * cdef char* f = NULL # <<<<<<<<<<<<<< + * cdef dtype descr = PyArray_DESCR(self) + * cdef int offset + */ + __pyx_v_f = NULL; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":297 + * cdef int t + * cdef char* f = NULL + * cdef dtype descr = PyArray_DESCR(self) # <<<<<<<<<<<<<< + * cdef int offset + * + */ + __pyx_t_7 = PyArray_DESCR(__pyx_v_self); + __pyx_t_3 = ((PyObject *)__pyx_t_7); + __Pyx_INCREF(__pyx_t_3); + __pyx_v_descr = ((PyArray_Descr *)__pyx_t_3); + __pyx_t_3 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":300 + * cdef int offset + * + * info.obj = self # <<<<<<<<<<<<<< + * + * if not PyDataType_HASFIELDS(descr): + */ + __Pyx_INCREF(((PyObject *)__pyx_v_self)); + __Pyx_GIVEREF(((PyObject *)__pyx_v_self)); + __Pyx_GOTREF(__pyx_v_info->obj); + __Pyx_DECREF(__pyx_v_info->obj); + __pyx_v_info->obj = ((PyObject *)__pyx_v_self); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":302 + * info.obj = self + * + * if not PyDataType_HASFIELDS(descr): # <<<<<<<<<<<<<< + * t = descr.type_num + * if ((descr.byteorder == c'>' and little_endian) or + */ + __pyx_t_1 = ((!(PyDataType_HASFIELDS(__pyx_v_descr) != 0)) != 0); + if (__pyx_t_1) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":303 + * + * if not PyDataType_HASFIELDS(descr): + * t = descr.type_num # <<<<<<<<<<<<<< + * if ((descr.byteorder == c'>' and little_endian) or + * (descr.byteorder == c'<' and not little_endian)): + */ + __pyx_t_4 = __pyx_v_descr->type_num; + __pyx_v_t = __pyx_t_4; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":304 + * if not PyDataType_HASFIELDS(descr): + * t = descr.type_num + * if ((descr.byteorder == c'>' and little_endian) or # <<<<<<<<<<<<<< + * (descr.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + */ + __pyx_t_2 = ((__pyx_v_descr->byteorder == '>') != 0); + if (!__pyx_t_2) { + goto __pyx_L15_next_or; + } else { + } + __pyx_t_2 = (__pyx_v_little_endian != 0); + if (!__pyx_t_2) { + } else { + __pyx_t_1 = __pyx_t_2; + goto __pyx_L14_bool_binop_done; + } + __pyx_L15_next_or:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":305 + * t = descr.type_num + * if ((descr.byteorder == c'>' and little_endian) or + * (descr.byteorder == c'<' and not little_endian)): # <<<<<<<<<<<<<< + * raise ValueError(u"Non-native byte order not supported") + * if t == NPY_BYTE: f = "b" + */ + __pyx_t_2 = ((__pyx_v_descr->byteorder == '<') != 0); + if (__pyx_t_2) { + } else { + __pyx_t_1 = __pyx_t_2; + goto __pyx_L14_bool_binop_done; + } + __pyx_t_2 = ((!(__pyx_v_little_endian != 0)) != 0); + __pyx_t_1 = __pyx_t_2; + __pyx_L14_bool_binop_done:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":304 + * if not PyDataType_HASFIELDS(descr): + * t = descr.type_num + * if ((descr.byteorder == c'>' and little_endian) or # <<<<<<<<<<<<<< + * (descr.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + */ + if (unlikely(__pyx_t_1)) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":306 + * if ((descr.byteorder == c'>' and little_endian) or + * (descr.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") # <<<<<<<<<<<<<< + * if t == NPY_BYTE: f = "b" + * elif t == NPY_UBYTE: f = "B" + */ + __pyx_t_3 = __Pyx_PyObject_Call(__pyx_builtin_ValueError, __pyx_tuple__3, NULL); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 306, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_Raise(__pyx_t_3, 0, 0, 0); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __PYX_ERR(1, 306, __pyx_L1_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":304 + * if not PyDataType_HASFIELDS(descr): + * t = descr.type_num + * if ((descr.byteorder == c'>' and little_endian) or # <<<<<<<<<<<<<< + * (descr.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":307 + * (descr.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + * if t == NPY_BYTE: f = "b" # <<<<<<<<<<<<<< + * elif t == NPY_UBYTE: f = "B" + * elif t == NPY_SHORT: f = "h" + */ + switch (__pyx_v_t) { + case NPY_BYTE: + __pyx_v_f = ((char *)"b"); + break; + case NPY_UBYTE: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":308 + * raise ValueError(u"Non-native byte order not supported") + * if t == NPY_BYTE: f = "b" + * elif t == NPY_UBYTE: f = "B" # <<<<<<<<<<<<<< + * elif t == NPY_SHORT: f = "h" + * elif t == NPY_USHORT: f = "H" + */ + __pyx_v_f = ((char *)"B"); + break; + case NPY_SHORT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":309 + * if t == NPY_BYTE: f = "b" + * elif t == NPY_UBYTE: f = "B" + * elif t == NPY_SHORT: f = "h" # <<<<<<<<<<<<<< + * elif t == NPY_USHORT: f = "H" + * elif t == NPY_INT: f = "i" + */ + __pyx_v_f = ((char *)"h"); + break; + case NPY_USHORT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":310 + * elif t == NPY_UBYTE: f = "B" + * elif t == NPY_SHORT: f = "h" + * elif t == NPY_USHORT: f = "H" # <<<<<<<<<<<<<< + * elif t == NPY_INT: f = "i" + * elif t == NPY_UINT: f = "I" + */ + __pyx_v_f = ((char *)"H"); + break; + case NPY_INT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":311 + * elif t == NPY_SHORT: f = "h" + * elif t == NPY_USHORT: f = "H" + * elif t == NPY_INT: f = "i" # <<<<<<<<<<<<<< + * elif t == NPY_UINT: f = "I" + * elif t == NPY_LONG: f = "l" + */ + __pyx_v_f = ((char *)"i"); + break; + case NPY_UINT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":312 + * elif t == NPY_USHORT: f = "H" + * elif t == NPY_INT: f = "i" + * elif t == NPY_UINT: f = "I" # <<<<<<<<<<<<<< + * elif t == NPY_LONG: f = "l" + * elif t == NPY_ULONG: f = "L" + */ + __pyx_v_f = ((char *)"I"); + break; + case NPY_LONG: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":313 + * elif t == NPY_INT: f = "i" + * elif t == NPY_UINT: f = "I" + * elif t == NPY_LONG: f = "l" # <<<<<<<<<<<<<< + * elif t == NPY_ULONG: f = "L" + * elif t == NPY_LONGLONG: f = "q" + */ + __pyx_v_f = ((char *)"l"); + break; + case NPY_ULONG: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":314 + * elif t == NPY_UINT: f = "I" + * elif t == NPY_LONG: f = "l" + * elif t == NPY_ULONG: f = "L" # <<<<<<<<<<<<<< + * elif t == NPY_LONGLONG: f = "q" + * elif t == NPY_ULONGLONG: f = "Q" + */ + __pyx_v_f = ((char *)"L"); + break; + case NPY_LONGLONG: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":315 + * elif t == NPY_LONG: f = "l" + * elif t == NPY_ULONG: f = "L" + * elif t == NPY_LONGLONG: f = "q" # <<<<<<<<<<<<<< + * elif t == NPY_ULONGLONG: f = "Q" + * elif t == NPY_FLOAT: f = "f" + */ + __pyx_v_f = ((char *)"q"); + break; + case NPY_ULONGLONG: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":316 + * elif t == NPY_ULONG: f = "L" + * elif t == NPY_LONGLONG: f = "q" + * elif t == NPY_ULONGLONG: f = "Q" # <<<<<<<<<<<<<< + * elif t == NPY_FLOAT: f = "f" + * elif t == NPY_DOUBLE: f = "d" + */ + __pyx_v_f = ((char *)"Q"); + break; + case NPY_FLOAT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":317 + * elif t == NPY_LONGLONG: f = "q" + * elif t == NPY_ULONGLONG: f = "Q" + * elif t == NPY_FLOAT: f = "f" # <<<<<<<<<<<<<< + * elif t == NPY_DOUBLE: f = "d" + * elif t == NPY_LONGDOUBLE: f = "g" + */ + __pyx_v_f = ((char *)"f"); + break; + case NPY_DOUBLE: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":318 + * elif t == NPY_ULONGLONG: f = "Q" + * elif t == NPY_FLOAT: f = "f" + * elif t == NPY_DOUBLE: f = "d" # <<<<<<<<<<<<<< + * elif t == NPY_LONGDOUBLE: f = "g" + * elif t == NPY_CFLOAT: f = "Zf" + */ + __pyx_v_f = ((char *)"d"); + break; + case NPY_LONGDOUBLE: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":319 + * elif t == NPY_FLOAT: f = "f" + * elif t == NPY_DOUBLE: f = "d" + * elif t == NPY_LONGDOUBLE: f = "g" # <<<<<<<<<<<<<< + * elif t == NPY_CFLOAT: f = "Zf" + * elif t == NPY_CDOUBLE: f = "Zd" + */ + __pyx_v_f = ((char *)"g"); + break; + case NPY_CFLOAT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":320 + * elif t == NPY_DOUBLE: f = "d" + * elif t == NPY_LONGDOUBLE: f = "g" + * elif t == NPY_CFLOAT: f = "Zf" # <<<<<<<<<<<<<< + * elif t == NPY_CDOUBLE: f = "Zd" + * elif t == NPY_CLONGDOUBLE: f = "Zg" + */ + __pyx_v_f = ((char *)"Zf"); + break; + case NPY_CDOUBLE: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":321 + * elif t == NPY_LONGDOUBLE: f = "g" + * elif t == NPY_CFLOAT: f = "Zf" + * elif t == NPY_CDOUBLE: f = "Zd" # <<<<<<<<<<<<<< + * elif t == NPY_CLONGDOUBLE: f = "Zg" + * elif t == NPY_OBJECT: f = "O" + */ + __pyx_v_f = ((char *)"Zd"); + break; + case NPY_CLONGDOUBLE: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":322 + * elif t == NPY_CFLOAT: f = "Zf" + * elif t == NPY_CDOUBLE: f = "Zd" + * elif t == NPY_CLONGDOUBLE: f = "Zg" # <<<<<<<<<<<<<< + * elif t == NPY_OBJECT: f = "O" + * else: + */ + __pyx_v_f = ((char *)"Zg"); + break; + case NPY_OBJECT: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":323 + * elif t == NPY_CDOUBLE: f = "Zd" + * elif t == NPY_CLONGDOUBLE: f = "Zg" + * elif t == NPY_OBJECT: f = "O" # <<<<<<<<<<<<<< + * else: + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) + */ + __pyx_v_f = ((char *)"O"); + break; + default: + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":325 + * elif t == NPY_OBJECT: f = "O" + * else: + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) # <<<<<<<<<<<<<< + * info.format = f + * return + */ + __pyx_t_3 = __Pyx_PyInt_From_int(__pyx_v_t); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 325, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_8 = PyUnicode_Format(__pyx_kp_u_unknown_dtype_code_in_numpy_pxd, __pyx_t_3); if (unlikely(!__pyx_t_8)) __PYX_ERR(1, 325, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_8); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_3 = __Pyx_PyObject_CallOneArg(__pyx_builtin_ValueError, __pyx_t_8); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 325, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; + __Pyx_Raise(__pyx_t_3, 0, 0, 0); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __PYX_ERR(1, 325, __pyx_L1_error) + break; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":326 + * else: + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) + * info.format = f # <<<<<<<<<<<<<< + * return + * else: + */ + __pyx_v_info->format = __pyx_v_f; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":327 + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) + * info.format = f + * return # <<<<<<<<<<<<<< + * else: + * info.format = PyObject_Malloc(_buffer_format_string_len) + */ + __pyx_r = 0; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":302 + * info.obj = self + * + * if not PyDataType_HASFIELDS(descr): # <<<<<<<<<<<<<< + * t = descr.type_num + * if ((descr.byteorder == c'>' and little_endian) or + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":329 + * return + * else: + * info.format = PyObject_Malloc(_buffer_format_string_len) # <<<<<<<<<<<<<< + * info.format[0] = c'^' # Native data types, manual alignment + * offset = 0 + */ + /*else*/ { + __pyx_v_info->format = ((char *)PyObject_Malloc(0xFF)); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":330 + * else: + * info.format = PyObject_Malloc(_buffer_format_string_len) + * info.format[0] = c'^' # Native data types, manual alignment # <<<<<<<<<<<<<< + * offset = 0 + * f = _util_dtypestring(descr, info.format + 1, + */ + (__pyx_v_info->format[0]) = '^'; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":331 + * info.format = PyObject_Malloc(_buffer_format_string_len) + * info.format[0] = c'^' # Native data types, manual alignment + * offset = 0 # <<<<<<<<<<<<<< + * f = _util_dtypestring(descr, info.format + 1, + * info.format + _buffer_format_string_len, + */ + __pyx_v_offset = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":332 + * info.format[0] = c'^' # Native data types, manual alignment + * offset = 0 + * f = _util_dtypestring(descr, info.format + 1, # <<<<<<<<<<<<<< + * info.format + _buffer_format_string_len, + * &offset) + */ + __pyx_t_9 = __pyx_f_5numpy__util_dtypestring(__pyx_v_descr, (__pyx_v_info->format + 1), (__pyx_v_info->format + 0xFF), (&__pyx_v_offset)); if (unlikely(__pyx_t_9 == ((char *)NULL))) __PYX_ERR(1, 332, __pyx_L1_error) + __pyx_v_f = __pyx_t_9; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":335 + * info.format + _buffer_format_string_len, + * &offset) + * f[0] = c'\0' # Terminate format string # <<<<<<<<<<<<<< + * + * def __releasebuffer__(ndarray self, Py_buffer* info): + */ + (__pyx_v_f[0]) = '\x00'; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":258 + * # experimental exception made for __getbuffer__ and __releasebuffer__ + * # -- the details of this may change. + * def __getbuffer__(ndarray self, Py_buffer* info, int flags): # <<<<<<<<<<<<<< + * # This implementation of getbuffer is geared towards Cython + * # requirements, and does not yet fulfill the PEP. + */ + + /* function exit code */ + __pyx_r = 0; + goto __pyx_L0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_3); + __Pyx_XDECREF(__pyx_t_8); + __Pyx_AddTraceback("numpy.ndarray.__getbuffer__", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = -1; + if (__pyx_v_info->obj != NULL) { + __Pyx_GOTREF(__pyx_v_info->obj); + __Pyx_DECREF(__pyx_v_info->obj); __pyx_v_info->obj = 0; + } + goto __pyx_L2; + __pyx_L0:; + if (__pyx_v_info->obj == Py_None) { + __Pyx_GOTREF(__pyx_v_info->obj); + __Pyx_DECREF(__pyx_v_info->obj); __pyx_v_info->obj = 0; + } + __pyx_L2:; + __Pyx_XDECREF((PyObject *)__pyx_v_descr); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":337 + * f[0] = c'\0' # Terminate format string + * + * def __releasebuffer__(ndarray self, Py_buffer* info): # <<<<<<<<<<<<<< + * if PyArray_HASFIELDS(self): + * PyObject_Free(info.format) + */ + +/* Python wrapper */ +static CYTHON_UNUSED void __pyx_pw_5numpy_7ndarray_3__releasebuffer__(PyObject *__pyx_v_self, Py_buffer *__pyx_v_info); /*proto*/ +static CYTHON_UNUSED void __pyx_pw_5numpy_7ndarray_3__releasebuffer__(PyObject *__pyx_v_self, Py_buffer *__pyx_v_info) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__releasebuffer__ (wrapper)", 0); + __pyx_pf_5numpy_7ndarray_2__releasebuffer__(((PyArrayObject *)__pyx_v_self), ((Py_buffer *)__pyx_v_info)); + + /* function exit code */ + __Pyx_RefNannyFinishContext(); +} + +static void __pyx_pf_5numpy_7ndarray_2__releasebuffer__(PyArrayObject *__pyx_v_self, Py_buffer *__pyx_v_info) { + __Pyx_RefNannyDeclarations + int __pyx_t_1; + __Pyx_RefNannySetupContext("__releasebuffer__", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":338 + * + * def __releasebuffer__(ndarray self, Py_buffer* info): + * if PyArray_HASFIELDS(self): # <<<<<<<<<<<<<< + * PyObject_Free(info.format) + * if sizeof(npy_intp) != sizeof(Py_ssize_t): + */ + __pyx_t_1 = (PyArray_HASFIELDS(__pyx_v_self) != 0); + if (__pyx_t_1) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":339 + * def __releasebuffer__(ndarray self, Py_buffer* info): + * if PyArray_HASFIELDS(self): + * PyObject_Free(info.format) # <<<<<<<<<<<<<< + * if sizeof(npy_intp) != sizeof(Py_ssize_t): + * PyObject_Free(info.strides) + */ + PyObject_Free(__pyx_v_info->format); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":338 + * + * def __releasebuffer__(ndarray self, Py_buffer* info): + * if PyArray_HASFIELDS(self): # <<<<<<<<<<<<<< + * PyObject_Free(info.format) + * if sizeof(npy_intp) != sizeof(Py_ssize_t): + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":340 + * if PyArray_HASFIELDS(self): + * PyObject_Free(info.format) + * if sizeof(npy_intp) != sizeof(Py_ssize_t): # <<<<<<<<<<<<<< + * PyObject_Free(info.strides) + * # info.shape was stored after info.strides in the same block + */ + __pyx_t_1 = (((sizeof(npy_intp)) != (sizeof(Py_ssize_t))) != 0); + if (__pyx_t_1) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":341 + * PyObject_Free(info.format) + * if sizeof(npy_intp) != sizeof(Py_ssize_t): + * PyObject_Free(info.strides) # <<<<<<<<<<<<<< + * # info.shape was stored after info.strides in the same block + * + */ + PyObject_Free(__pyx_v_info->strides); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":340 + * if PyArray_HASFIELDS(self): + * PyObject_Free(info.format) + * if sizeof(npy_intp) != sizeof(Py_ssize_t): # <<<<<<<<<<<<<< + * PyObject_Free(info.strides) + * # info.shape was stored after info.strides in the same block + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":337 + * f[0] = c'\0' # Terminate format string + * + * def __releasebuffer__(ndarray self, Py_buffer* info): # <<<<<<<<<<<<<< + * if PyArray_HASFIELDS(self): + * PyObject_Free(info.format) + */ + + /* function exit code */ + __Pyx_RefNannyFinishContext(); +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":821 + * ctypedef npy_cdouble complex_t + * + * cdef inline object PyArray_MultiIterNew1(a): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(1, a) + * + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_PyArray_MultiIterNew1(PyObject *__pyx_v_a) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("PyArray_MultiIterNew1", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":822 + * + * cdef inline object PyArray_MultiIterNew1(a): + * return PyArray_MultiIterNew(1, a) # <<<<<<<<<<<<<< + * + * cdef inline object PyArray_MultiIterNew2(a, b): + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = PyArray_MultiIterNew(1, ((void *)__pyx_v_a)); if (unlikely(!__pyx_t_1)) __PYX_ERR(1, 822, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_r = __pyx_t_1; + __pyx_t_1 = 0; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":821 + * ctypedef npy_cdouble complex_t + * + * cdef inline object PyArray_MultiIterNew1(a): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(1, a) + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_AddTraceback("numpy.PyArray_MultiIterNew1", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":824 + * return PyArray_MultiIterNew(1, a) + * + * cdef inline object PyArray_MultiIterNew2(a, b): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(2, a, b) + * + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_PyArray_MultiIterNew2(PyObject *__pyx_v_a, PyObject *__pyx_v_b) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("PyArray_MultiIterNew2", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":825 + * + * cdef inline object PyArray_MultiIterNew2(a, b): + * return PyArray_MultiIterNew(2, a, b) # <<<<<<<<<<<<<< + * + * cdef inline object PyArray_MultiIterNew3(a, b, c): + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = PyArray_MultiIterNew(2, ((void *)__pyx_v_a), ((void *)__pyx_v_b)); if (unlikely(!__pyx_t_1)) __PYX_ERR(1, 825, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_r = __pyx_t_1; + __pyx_t_1 = 0; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":824 + * return PyArray_MultiIterNew(1, a) + * + * cdef inline object PyArray_MultiIterNew2(a, b): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(2, a, b) + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_AddTraceback("numpy.PyArray_MultiIterNew2", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":827 + * return PyArray_MultiIterNew(2, a, b) + * + * cdef inline object PyArray_MultiIterNew3(a, b, c): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(3, a, b, c) + * + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_PyArray_MultiIterNew3(PyObject *__pyx_v_a, PyObject *__pyx_v_b, PyObject *__pyx_v_c) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("PyArray_MultiIterNew3", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":828 + * + * cdef inline object PyArray_MultiIterNew3(a, b, c): + * return PyArray_MultiIterNew(3, a, b, c) # <<<<<<<<<<<<<< + * + * cdef inline object PyArray_MultiIterNew4(a, b, c, d): + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = PyArray_MultiIterNew(3, ((void *)__pyx_v_a), ((void *)__pyx_v_b), ((void *)__pyx_v_c)); if (unlikely(!__pyx_t_1)) __PYX_ERR(1, 828, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_r = __pyx_t_1; + __pyx_t_1 = 0; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":827 + * return PyArray_MultiIterNew(2, a, b) + * + * cdef inline object PyArray_MultiIterNew3(a, b, c): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(3, a, b, c) + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_AddTraceback("numpy.PyArray_MultiIterNew3", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":830 + * return PyArray_MultiIterNew(3, a, b, c) + * + * cdef inline object PyArray_MultiIterNew4(a, b, c, d): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(4, a, b, c, d) + * + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_PyArray_MultiIterNew4(PyObject *__pyx_v_a, PyObject *__pyx_v_b, PyObject *__pyx_v_c, PyObject *__pyx_v_d) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("PyArray_MultiIterNew4", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":831 + * + * cdef inline object PyArray_MultiIterNew4(a, b, c, d): + * return PyArray_MultiIterNew(4, a, b, c, d) # <<<<<<<<<<<<<< + * + * cdef inline object PyArray_MultiIterNew5(a, b, c, d, e): + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = PyArray_MultiIterNew(4, ((void *)__pyx_v_a), ((void *)__pyx_v_b), ((void *)__pyx_v_c), ((void *)__pyx_v_d)); if (unlikely(!__pyx_t_1)) __PYX_ERR(1, 831, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_r = __pyx_t_1; + __pyx_t_1 = 0; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":830 + * return PyArray_MultiIterNew(3, a, b, c) + * + * cdef inline object PyArray_MultiIterNew4(a, b, c, d): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(4, a, b, c, d) + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_AddTraceback("numpy.PyArray_MultiIterNew4", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":833 + * return PyArray_MultiIterNew(4, a, b, c, d) + * + * cdef inline object PyArray_MultiIterNew5(a, b, c, d, e): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(5, a, b, c, d, e) + * + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_PyArray_MultiIterNew5(PyObject *__pyx_v_a, PyObject *__pyx_v_b, PyObject *__pyx_v_c, PyObject *__pyx_v_d, PyObject *__pyx_v_e) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("PyArray_MultiIterNew5", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":834 + * + * cdef inline object PyArray_MultiIterNew5(a, b, c, d, e): + * return PyArray_MultiIterNew(5, a, b, c, d, e) # <<<<<<<<<<<<<< + * + * cdef inline tuple PyDataType_SHAPE(dtype d): + */ + __Pyx_XDECREF(__pyx_r); + __pyx_t_1 = PyArray_MultiIterNew(5, ((void *)__pyx_v_a), ((void *)__pyx_v_b), ((void *)__pyx_v_c), ((void *)__pyx_v_d), ((void *)__pyx_v_e)); if (unlikely(!__pyx_t_1)) __PYX_ERR(1, 834, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_r = __pyx_t_1; + __pyx_t_1 = 0; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":833 + * return PyArray_MultiIterNew(4, a, b, c, d) + * + * cdef inline object PyArray_MultiIterNew5(a, b, c, d, e): # <<<<<<<<<<<<<< + * return PyArray_MultiIterNew(5, a, b, c, d, e) + * + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_AddTraceback("numpy.PyArray_MultiIterNew5", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = 0; + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":836 + * return PyArray_MultiIterNew(5, a, b, c, d, e) + * + * cdef inline tuple PyDataType_SHAPE(dtype d): # <<<<<<<<<<<<<< + * if PyDataType_HASSUBARRAY(d): + * return d.subarray.shape + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_PyDataType_SHAPE(PyArray_Descr *__pyx_v_d) { + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + int __pyx_t_1; + __Pyx_RefNannySetupContext("PyDataType_SHAPE", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":837 + * + * cdef inline tuple PyDataType_SHAPE(dtype d): + * if PyDataType_HASSUBARRAY(d): # <<<<<<<<<<<<<< + * return d.subarray.shape + * else: + */ + __pyx_t_1 = (PyDataType_HASSUBARRAY(__pyx_v_d) != 0); + if (__pyx_t_1) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":838 + * cdef inline tuple PyDataType_SHAPE(dtype d): + * if PyDataType_HASSUBARRAY(d): + * return d.subarray.shape # <<<<<<<<<<<<<< + * else: + * return () + */ + __Pyx_XDECREF(__pyx_r); + __Pyx_INCREF(((PyObject*)__pyx_v_d->subarray->shape)); + __pyx_r = ((PyObject*)__pyx_v_d->subarray->shape); + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":837 + * + * cdef inline tuple PyDataType_SHAPE(dtype d): + * if PyDataType_HASSUBARRAY(d): # <<<<<<<<<<<<<< + * return d.subarray.shape + * else: + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":840 + * return d.subarray.shape + * else: + * return () # <<<<<<<<<<<<<< + * + * cdef inline char* _util_dtypestring(dtype descr, char* f, char* end, int* offset) except NULL: + */ + /*else*/ { + __Pyx_XDECREF(__pyx_r); + __Pyx_INCREF(__pyx_empty_tuple); + __pyx_r = __pyx_empty_tuple; + goto __pyx_L0; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":836 + * return PyArray_MultiIterNew(5, a, b, c, d, e) + * + * cdef inline tuple PyDataType_SHAPE(dtype d): # <<<<<<<<<<<<<< + * if PyDataType_HASSUBARRAY(d): + * return d.subarray.shape + */ + + /* function exit code */ + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":842 + * return () + * + * cdef inline char* _util_dtypestring(dtype descr, char* f, char* end, int* offset) except NULL: # <<<<<<<<<<<<<< + * # Recursive utility function used in __getbuffer__ to get format + * # string. The new location in the format string is returned. + */ + +static CYTHON_INLINE char *__pyx_f_5numpy__util_dtypestring(PyArray_Descr *__pyx_v_descr, char *__pyx_v_f, char *__pyx_v_end, int *__pyx_v_offset) { + PyArray_Descr *__pyx_v_child = 0; + int __pyx_v_endian_detector; + int __pyx_v_little_endian; + PyObject *__pyx_v_fields = 0; + PyObject *__pyx_v_childname = NULL; + PyObject *__pyx_v_new_offset = NULL; + PyObject *__pyx_v_t = NULL; + char *__pyx_r; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + Py_ssize_t __pyx_t_2; + PyObject *__pyx_t_3 = NULL; + PyObject *__pyx_t_4 = NULL; + int __pyx_t_5; + int __pyx_t_6; + int __pyx_t_7; + long __pyx_t_8; + char *__pyx_t_9; + __Pyx_RefNannySetupContext("_util_dtypestring", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":847 + * + * cdef dtype child + * cdef int endian_detector = 1 # <<<<<<<<<<<<<< + * cdef bint little_endian = ((&endian_detector)[0] != 0) + * cdef tuple fields + */ + __pyx_v_endian_detector = 1; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":848 + * cdef dtype child + * cdef int endian_detector = 1 + * cdef bint little_endian = ((&endian_detector)[0] != 0) # <<<<<<<<<<<<<< + * cdef tuple fields + * + */ + __pyx_v_little_endian = ((((char *)(&__pyx_v_endian_detector))[0]) != 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":851 + * cdef tuple fields + * + * for childname in descr.names: # <<<<<<<<<<<<<< + * fields = descr.fields[childname] + * child, new_offset = fields + */ + if (unlikely(__pyx_v_descr->names == Py_None)) { + PyErr_SetString(PyExc_TypeError, "'NoneType' object is not iterable"); + __PYX_ERR(1, 851, __pyx_L1_error) + } + __pyx_t_1 = __pyx_v_descr->names; __Pyx_INCREF(__pyx_t_1); __pyx_t_2 = 0; + for (;;) { + if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_1)) break; + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_3 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_3); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(1, 851, __pyx_L1_error) + #else + __pyx_t_3 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 851, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + #endif + __Pyx_XDECREF_SET(__pyx_v_childname, __pyx_t_3); + __pyx_t_3 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":852 + * + * for childname in descr.names: + * fields = descr.fields[childname] # <<<<<<<<<<<<<< + * child, new_offset = fields + * + */ + if (unlikely(__pyx_v_descr->fields == Py_None)) { + PyErr_SetString(PyExc_TypeError, "'NoneType' object is not subscriptable"); + __PYX_ERR(1, 852, __pyx_L1_error) + } + __pyx_t_3 = __Pyx_PyDict_GetItem(__pyx_v_descr->fields, __pyx_v_childname); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 852, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + if (!(likely(PyTuple_CheckExact(__pyx_t_3))||((__pyx_t_3) == Py_None)||(PyErr_Format(PyExc_TypeError, "Expected %.16s, got %.200s", "tuple", Py_TYPE(__pyx_t_3)->tp_name), 0))) __PYX_ERR(1, 852, __pyx_L1_error) + __Pyx_XDECREF_SET(__pyx_v_fields, ((PyObject*)__pyx_t_3)); + __pyx_t_3 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":853 + * for childname in descr.names: + * fields = descr.fields[childname] + * child, new_offset = fields # <<<<<<<<<<<<<< + * + * if (end - f) - (new_offset - offset[0]) < 15: + */ + if (likely(__pyx_v_fields != Py_None)) { + PyObject* sequence = __pyx_v_fields; + Py_ssize_t size = __Pyx_PySequence_SIZE(sequence); + if (unlikely(size != 2)) { + if (size > 2) __Pyx_RaiseTooManyValuesError(2); + else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); + __PYX_ERR(1, 853, __pyx_L1_error) + } + #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS + __pyx_t_3 = PyTuple_GET_ITEM(sequence, 0); + __pyx_t_4 = PyTuple_GET_ITEM(sequence, 1); + __Pyx_INCREF(__pyx_t_3); + __Pyx_INCREF(__pyx_t_4); + #else + __pyx_t_3 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 853, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 853, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + #endif + } else { + __Pyx_RaiseNoneNotIterableError(); __PYX_ERR(1, 853, __pyx_L1_error) + } + if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_dtype))))) __PYX_ERR(1, 853, __pyx_L1_error) + __Pyx_XDECREF_SET(__pyx_v_child, ((PyArray_Descr *)__pyx_t_3)); + __pyx_t_3 = 0; + __Pyx_XDECREF_SET(__pyx_v_new_offset, __pyx_t_4); + __pyx_t_4 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":855 + * child, new_offset = fields + * + * if (end - f) - (new_offset - offset[0]) < 15: # <<<<<<<<<<<<<< + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") + * + */ + __pyx_t_4 = __Pyx_PyInt_From_int((__pyx_v_offset[0])); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 855, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyNumber_Subtract(__pyx_v_new_offset, __pyx_t_4); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 855, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_5 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_5 == (int)-1) && PyErr_Occurred())) __PYX_ERR(1, 855, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = ((((__pyx_v_end - __pyx_v_f) - ((int)__pyx_t_5)) < 15) != 0); + if (unlikely(__pyx_t_6)) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":856 + * + * if (end - f) - (new_offset - offset[0]) < 15: + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") # <<<<<<<<<<<<<< + * + * if ((child.byteorder == c'>' and little_endian) or + */ + __pyx_t_3 = __Pyx_PyObject_Call(__pyx_builtin_RuntimeError, __pyx_tuple__4, NULL); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 856, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_Raise(__pyx_t_3, 0, 0, 0); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __PYX_ERR(1, 856, __pyx_L1_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":855 + * child, new_offset = fields + * + * if (end - f) - (new_offset - offset[0]) < 15: # <<<<<<<<<<<<<< + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") + * + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":858 + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") + * + * if ((child.byteorder == c'>' and little_endian) or # <<<<<<<<<<<<<< + * (child.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + */ + __pyx_t_7 = ((__pyx_v_child->byteorder == '>') != 0); + if (!__pyx_t_7) { + goto __pyx_L8_next_or; + } else { + } + __pyx_t_7 = (__pyx_v_little_endian != 0); + if (!__pyx_t_7) { + } else { + __pyx_t_6 = __pyx_t_7; + goto __pyx_L7_bool_binop_done; + } + __pyx_L8_next_or:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":859 + * + * if ((child.byteorder == c'>' and little_endian) or + * (child.byteorder == c'<' and not little_endian)): # <<<<<<<<<<<<<< + * raise ValueError(u"Non-native byte order not supported") + * # One could encode it in the format string and have Cython + */ + __pyx_t_7 = ((__pyx_v_child->byteorder == '<') != 0); + if (__pyx_t_7) { + } else { + __pyx_t_6 = __pyx_t_7; + goto __pyx_L7_bool_binop_done; + } + __pyx_t_7 = ((!(__pyx_v_little_endian != 0)) != 0); + __pyx_t_6 = __pyx_t_7; + __pyx_L7_bool_binop_done:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":858 + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") + * + * if ((child.byteorder == c'>' and little_endian) or # <<<<<<<<<<<<<< + * (child.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + */ + if (unlikely(__pyx_t_6)) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":860 + * if ((child.byteorder == c'>' and little_endian) or + * (child.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") # <<<<<<<<<<<<<< + * # One could encode it in the format string and have Cython + * # complain instead, BUT: < and > in format strings also imply + */ + __pyx_t_3 = __Pyx_PyObject_Call(__pyx_builtin_ValueError, __pyx_tuple__3, NULL); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 860, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __Pyx_Raise(__pyx_t_3, 0, 0, 0); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __PYX_ERR(1, 860, __pyx_L1_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":858 + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") + * + * if ((child.byteorder == c'>' and little_endian) or # <<<<<<<<<<<<<< + * (child.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":870 + * + * # Output padding bytes + * while offset[0] < new_offset: # <<<<<<<<<<<<<< + * f[0] = 120 # "x"; pad byte + * f += 1 + */ + while (1) { + __pyx_t_3 = __Pyx_PyInt_From_int((__pyx_v_offset[0])); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 870, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_t_3, __pyx_v_new_offset, Py_LT); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 870, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 870, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (!__pyx_t_6) break; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":871 + * # Output padding bytes + * while offset[0] < new_offset: + * f[0] = 120 # "x"; pad byte # <<<<<<<<<<<<<< + * f += 1 + * offset[0] += 1 + */ + (__pyx_v_f[0]) = 0x78; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":872 + * while offset[0] < new_offset: + * f[0] = 120 # "x"; pad byte + * f += 1 # <<<<<<<<<<<<<< + * offset[0] += 1 + * + */ + __pyx_v_f = (__pyx_v_f + 1); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":873 + * f[0] = 120 # "x"; pad byte + * f += 1 + * offset[0] += 1 # <<<<<<<<<<<<<< + * + * offset[0] += child.itemsize + */ + __pyx_t_8 = 0; + (__pyx_v_offset[__pyx_t_8]) = ((__pyx_v_offset[__pyx_t_8]) + 1); + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":875 + * offset[0] += 1 + * + * offset[0] += child.itemsize # <<<<<<<<<<<<<< + * + * if not PyDataType_HASFIELDS(child): + */ + __pyx_t_8 = 0; + (__pyx_v_offset[__pyx_t_8]) = ((__pyx_v_offset[__pyx_t_8]) + __pyx_v_child->elsize); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":877 + * offset[0] += child.itemsize + * + * if not PyDataType_HASFIELDS(child): # <<<<<<<<<<<<<< + * t = child.type_num + * if end - f < 5: + */ + __pyx_t_6 = ((!(PyDataType_HASFIELDS(__pyx_v_child) != 0)) != 0); + if (__pyx_t_6) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":878 + * + * if not PyDataType_HASFIELDS(child): + * t = child.type_num # <<<<<<<<<<<<<< + * if end - f < 5: + * raise RuntimeError(u"Format string allocated too short.") + */ + __pyx_t_4 = __Pyx_PyInt_From_int(__pyx_v_child->type_num); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 878, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __Pyx_XDECREF_SET(__pyx_v_t, __pyx_t_4); + __pyx_t_4 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":879 + * if not PyDataType_HASFIELDS(child): + * t = child.type_num + * if end - f < 5: # <<<<<<<<<<<<<< + * raise RuntimeError(u"Format string allocated too short.") + * + */ + __pyx_t_6 = (((__pyx_v_end - __pyx_v_f) < 5) != 0); + if (unlikely(__pyx_t_6)) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":880 + * t = child.type_num + * if end - f < 5: + * raise RuntimeError(u"Format string allocated too short.") # <<<<<<<<<<<<<< + * + * # Until ticket #99 is fixed, use integers to avoid warnings + */ + __pyx_t_4 = __Pyx_PyObject_Call(__pyx_builtin_RuntimeError, __pyx_tuple__5, NULL); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 880, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __Pyx_Raise(__pyx_t_4, 0, 0, 0); + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __PYX_ERR(1, 880, __pyx_L1_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":879 + * if not PyDataType_HASFIELDS(child): + * t = child.type_num + * if end - f < 5: # <<<<<<<<<<<<<< + * raise RuntimeError(u"Format string allocated too short.") + * + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":883 + * + * # Until ticket #99 is fixed, use integers to avoid warnings + * if t == NPY_BYTE: f[0] = 98 #"b" # <<<<<<<<<<<<<< + * elif t == NPY_UBYTE: f[0] = 66 #"B" + * elif t == NPY_SHORT: f[0] = 104 #"h" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_BYTE); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 883, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 883, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 883, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 98; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":884 + * # Until ticket #99 is fixed, use integers to avoid warnings + * if t == NPY_BYTE: f[0] = 98 #"b" + * elif t == NPY_UBYTE: f[0] = 66 #"B" # <<<<<<<<<<<<<< + * elif t == NPY_SHORT: f[0] = 104 #"h" + * elif t == NPY_USHORT: f[0] = 72 #"H" + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_UBYTE); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 884, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 884, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 884, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 66; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":885 + * if t == NPY_BYTE: f[0] = 98 #"b" + * elif t == NPY_UBYTE: f[0] = 66 #"B" + * elif t == NPY_SHORT: f[0] = 104 #"h" # <<<<<<<<<<<<<< + * elif t == NPY_USHORT: f[0] = 72 #"H" + * elif t == NPY_INT: f[0] = 105 #"i" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_SHORT); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 885, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 885, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 885, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x68; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":886 + * elif t == NPY_UBYTE: f[0] = 66 #"B" + * elif t == NPY_SHORT: f[0] = 104 #"h" + * elif t == NPY_USHORT: f[0] = 72 #"H" # <<<<<<<<<<<<<< + * elif t == NPY_INT: f[0] = 105 #"i" + * elif t == NPY_UINT: f[0] = 73 #"I" + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_USHORT); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 886, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 886, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 886, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 72; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":887 + * elif t == NPY_SHORT: f[0] = 104 #"h" + * elif t == NPY_USHORT: f[0] = 72 #"H" + * elif t == NPY_INT: f[0] = 105 #"i" # <<<<<<<<<<<<<< + * elif t == NPY_UINT: f[0] = 73 #"I" + * elif t == NPY_LONG: f[0] = 108 #"l" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_INT); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 887, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 887, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 887, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x69; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":888 + * elif t == NPY_USHORT: f[0] = 72 #"H" + * elif t == NPY_INT: f[0] = 105 #"i" + * elif t == NPY_UINT: f[0] = 73 #"I" # <<<<<<<<<<<<<< + * elif t == NPY_LONG: f[0] = 108 #"l" + * elif t == NPY_ULONG: f[0] = 76 #"L" + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_UINT); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 888, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 888, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 888, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 73; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":889 + * elif t == NPY_INT: f[0] = 105 #"i" + * elif t == NPY_UINT: f[0] = 73 #"I" + * elif t == NPY_LONG: f[0] = 108 #"l" # <<<<<<<<<<<<<< + * elif t == NPY_ULONG: f[0] = 76 #"L" + * elif t == NPY_LONGLONG: f[0] = 113 #"q" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_LONG); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 889, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 889, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 889, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x6C; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":890 + * elif t == NPY_UINT: f[0] = 73 #"I" + * elif t == NPY_LONG: f[0] = 108 #"l" + * elif t == NPY_ULONG: f[0] = 76 #"L" # <<<<<<<<<<<<<< + * elif t == NPY_LONGLONG: f[0] = 113 #"q" + * elif t == NPY_ULONGLONG: f[0] = 81 #"Q" + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_ULONG); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 890, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 890, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 890, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 76; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":891 + * elif t == NPY_LONG: f[0] = 108 #"l" + * elif t == NPY_ULONG: f[0] = 76 #"L" + * elif t == NPY_LONGLONG: f[0] = 113 #"q" # <<<<<<<<<<<<<< + * elif t == NPY_ULONGLONG: f[0] = 81 #"Q" + * elif t == NPY_FLOAT: f[0] = 102 #"f" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_LONGLONG); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 891, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 891, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 891, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x71; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":892 + * elif t == NPY_ULONG: f[0] = 76 #"L" + * elif t == NPY_LONGLONG: f[0] = 113 #"q" + * elif t == NPY_ULONGLONG: f[0] = 81 #"Q" # <<<<<<<<<<<<<< + * elif t == NPY_FLOAT: f[0] = 102 #"f" + * elif t == NPY_DOUBLE: f[0] = 100 #"d" + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_ULONGLONG); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 892, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 892, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 892, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 81; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":893 + * elif t == NPY_LONGLONG: f[0] = 113 #"q" + * elif t == NPY_ULONGLONG: f[0] = 81 #"Q" + * elif t == NPY_FLOAT: f[0] = 102 #"f" # <<<<<<<<<<<<<< + * elif t == NPY_DOUBLE: f[0] = 100 #"d" + * elif t == NPY_LONGDOUBLE: f[0] = 103 #"g" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_FLOAT); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 893, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 893, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 893, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x66; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":894 + * elif t == NPY_ULONGLONG: f[0] = 81 #"Q" + * elif t == NPY_FLOAT: f[0] = 102 #"f" + * elif t == NPY_DOUBLE: f[0] = 100 #"d" # <<<<<<<<<<<<<< + * elif t == NPY_LONGDOUBLE: f[0] = 103 #"g" + * elif t == NPY_CFLOAT: f[0] = 90; f[1] = 102; f += 1 # Zf + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_DOUBLE); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 894, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 894, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 894, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x64; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":895 + * elif t == NPY_FLOAT: f[0] = 102 #"f" + * elif t == NPY_DOUBLE: f[0] = 100 #"d" + * elif t == NPY_LONGDOUBLE: f[0] = 103 #"g" # <<<<<<<<<<<<<< + * elif t == NPY_CFLOAT: f[0] = 90; f[1] = 102; f += 1 # Zf + * elif t == NPY_CDOUBLE: f[0] = 90; f[1] = 100; f += 1 # Zd + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_LONGDOUBLE); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 895, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 895, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 895, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 0x67; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":896 + * elif t == NPY_DOUBLE: f[0] = 100 #"d" + * elif t == NPY_LONGDOUBLE: f[0] = 103 #"g" + * elif t == NPY_CFLOAT: f[0] = 90; f[1] = 102; f += 1 # Zf # <<<<<<<<<<<<<< + * elif t == NPY_CDOUBLE: f[0] = 90; f[1] = 100; f += 1 # Zd + * elif t == NPY_CLONGDOUBLE: f[0] = 90; f[1] = 103; f += 1 # Zg + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_CFLOAT); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 896, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 896, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 896, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 90; + (__pyx_v_f[1]) = 0x66; + __pyx_v_f = (__pyx_v_f + 1); + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":897 + * elif t == NPY_LONGDOUBLE: f[0] = 103 #"g" + * elif t == NPY_CFLOAT: f[0] = 90; f[1] = 102; f += 1 # Zf + * elif t == NPY_CDOUBLE: f[0] = 90; f[1] = 100; f += 1 # Zd # <<<<<<<<<<<<<< + * elif t == NPY_CLONGDOUBLE: f[0] = 90; f[1] = 103; f += 1 # Zg + * elif t == NPY_OBJECT: f[0] = 79 #"O" + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_CDOUBLE); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 897, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 897, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 897, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 90; + (__pyx_v_f[1]) = 0x64; + __pyx_v_f = (__pyx_v_f + 1); + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":898 + * elif t == NPY_CFLOAT: f[0] = 90; f[1] = 102; f += 1 # Zf + * elif t == NPY_CDOUBLE: f[0] = 90; f[1] = 100; f += 1 # Zd + * elif t == NPY_CLONGDOUBLE: f[0] = 90; f[1] = 103; f += 1 # Zg # <<<<<<<<<<<<<< + * elif t == NPY_OBJECT: f[0] = 79 #"O" + * else: + */ + __pyx_t_3 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_CLONGDOUBLE); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 898, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = PyObject_RichCompare(__pyx_v_t, __pyx_t_3, Py_EQ); __Pyx_XGOTREF(__pyx_t_4); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 898, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 898, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + if (__pyx_t_6) { + (__pyx_v_f[0]) = 90; + (__pyx_v_f[1]) = 0x67; + __pyx_v_f = (__pyx_v_f + 1); + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":899 + * elif t == NPY_CDOUBLE: f[0] = 90; f[1] = 100; f += 1 # Zd + * elif t == NPY_CLONGDOUBLE: f[0] = 90; f[1] = 103; f += 1 # Zg + * elif t == NPY_OBJECT: f[0] = 79 #"O" # <<<<<<<<<<<<<< + * else: + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) + */ + __pyx_t_4 = __Pyx_PyInt_From_enum__NPY_TYPES(NPY_OBJECT); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 899, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __pyx_t_3 = PyObject_RichCompare(__pyx_v_t, __pyx_t_4, Py_EQ); __Pyx_XGOTREF(__pyx_t_3); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 899, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(1, 899, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + if (likely(__pyx_t_6)) { + (__pyx_v_f[0]) = 79; + goto __pyx_L15; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":901 + * elif t == NPY_OBJECT: f[0] = 79 #"O" + * else: + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) # <<<<<<<<<<<<<< + * f += 1 + * else: + */ + /*else*/ { + __pyx_t_3 = __Pyx_PyUnicode_FormatSafe(__pyx_kp_u_unknown_dtype_code_in_numpy_pxd, __pyx_v_t); if (unlikely(!__pyx_t_3)) __PYX_ERR(1, 901, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_3); + __pyx_t_4 = __Pyx_PyObject_CallOneArg(__pyx_builtin_ValueError, __pyx_t_3); if (unlikely(!__pyx_t_4)) __PYX_ERR(1, 901, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_4); + __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; + __Pyx_Raise(__pyx_t_4, 0, 0, 0); + __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __PYX_ERR(1, 901, __pyx_L1_error) + } + __pyx_L15:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":902 + * else: + * raise ValueError(u"unknown dtype code in numpy.pxd (%d)" % t) + * f += 1 # <<<<<<<<<<<<<< + * else: + * # Cython ignores struct boundary information ("T{...}"), + */ + __pyx_v_f = (__pyx_v_f + 1); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":877 + * offset[0] += child.itemsize + * + * if not PyDataType_HASFIELDS(child): # <<<<<<<<<<<<<< + * t = child.type_num + * if end - f < 5: + */ + goto __pyx_L13; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":906 + * # Cython ignores struct boundary information ("T{...}"), + * # so don't output it + * f = _util_dtypestring(child, f, end, offset) # <<<<<<<<<<<<<< + * return f + * + */ + /*else*/ { + __pyx_t_9 = __pyx_f_5numpy__util_dtypestring(__pyx_v_child, __pyx_v_f, __pyx_v_end, __pyx_v_offset); if (unlikely(__pyx_t_9 == ((char *)NULL))) __PYX_ERR(1, 906, __pyx_L1_error) + __pyx_v_f = __pyx_t_9; + } + __pyx_L13:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":851 + * cdef tuple fields + * + * for childname in descr.names: # <<<<<<<<<<<<<< + * fields = descr.fields[childname] + * child, new_offset = fields + */ + } + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":907 + * # so don't output it + * f = _util_dtypestring(child, f, end, offset) + * return f # <<<<<<<<<<<<<< + * + * + */ + __pyx_r = __pyx_v_f; + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":842 + * return () + * + * cdef inline char* _util_dtypestring(dtype descr, char* f, char* end, int* offset) except NULL: # <<<<<<<<<<<<<< + * # Recursive utility function used in __getbuffer__ to get format + * # string. The new location in the format string is returned. + */ + + /* function exit code */ + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_XDECREF(__pyx_t_3); + __Pyx_XDECREF(__pyx_t_4); + __Pyx_AddTraceback("numpy._util_dtypestring", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = NULL; + __pyx_L0:; + __Pyx_XDECREF((PyObject *)__pyx_v_child); + __Pyx_XDECREF(__pyx_v_fields); + __Pyx_XDECREF(__pyx_v_childname); + __Pyx_XDECREF(__pyx_v_new_offset); + __Pyx_XDECREF(__pyx_v_t); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1022 + * int _import_umath() except -1 + * + * cdef inline void set_array_base(ndarray arr, object base): # <<<<<<<<<<<<<< + * Py_INCREF(base) # important to do this before stealing the reference below! + * PyArray_SetBaseObject(arr, base) + */ + +static CYTHON_INLINE void __pyx_f_5numpy_set_array_base(PyArrayObject *__pyx_v_arr, PyObject *__pyx_v_base) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("set_array_base", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1023 + * + * cdef inline void set_array_base(ndarray arr, object base): + * Py_INCREF(base) # important to do this before stealing the reference below! # <<<<<<<<<<<<<< + * PyArray_SetBaseObject(arr, base) + * + */ + Py_INCREF(__pyx_v_base); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1024 + * cdef inline void set_array_base(ndarray arr, object base): + * Py_INCREF(base) # important to do this before stealing the reference below! + * PyArray_SetBaseObject(arr, base) # <<<<<<<<<<<<<< + * + * cdef inline object get_array_base(ndarray arr): + */ + (void)(PyArray_SetBaseObject(__pyx_v_arr, __pyx_v_base)); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1022 + * int _import_umath() except -1 + * + * cdef inline void set_array_base(ndarray arr, object base): # <<<<<<<<<<<<<< + * Py_INCREF(base) # important to do this before stealing the reference below! + * PyArray_SetBaseObject(arr, base) + */ + + /* function exit code */ + __Pyx_RefNannyFinishContext(); +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1026 + * PyArray_SetBaseObject(arr, base) + * + * cdef inline object get_array_base(ndarray arr): # <<<<<<<<<<<<<< + * base = PyArray_BASE(arr) + * if base is NULL: + */ + +static CYTHON_INLINE PyObject *__pyx_f_5numpy_get_array_base(PyArrayObject *__pyx_v_arr) { + PyObject *__pyx_v_base; + PyObject *__pyx_r = NULL; + __Pyx_RefNannyDeclarations + int __pyx_t_1; + __Pyx_RefNannySetupContext("get_array_base", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1027 + * + * cdef inline object get_array_base(ndarray arr): + * base = PyArray_BASE(arr) # <<<<<<<<<<<<<< + * if base is NULL: + * return None + */ + __pyx_v_base = PyArray_BASE(__pyx_v_arr); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1028 + * cdef inline object get_array_base(ndarray arr): + * base = PyArray_BASE(arr) + * if base is NULL: # <<<<<<<<<<<<<< + * return None + * return base + */ + __pyx_t_1 = ((__pyx_v_base == NULL) != 0); + if (__pyx_t_1) { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1029 + * base = PyArray_BASE(arr) + * if base is NULL: + * return None # <<<<<<<<<<<<<< + * return base + * + */ + __Pyx_XDECREF(__pyx_r); + __pyx_r = Py_None; __Pyx_INCREF(Py_None); + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1028 + * cdef inline object get_array_base(ndarray arr): + * base = PyArray_BASE(arr) + * if base is NULL: # <<<<<<<<<<<<<< + * return None + * return base + */ + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1030 + * if base is NULL: + * return None + * return base # <<<<<<<<<<<<<< + * + * # Versions of the import_* functions which are more suitable for + */ + __Pyx_XDECREF(__pyx_r); + __Pyx_INCREF(((PyObject *)__pyx_v_base)); + __pyx_r = ((PyObject *)__pyx_v_base); + goto __pyx_L0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1026 + * PyArray_SetBaseObject(arr, base) + * + * cdef inline object get_array_base(ndarray arr): # <<<<<<<<<<<<<< + * base = PyArray_BASE(arr) + * if base is NULL: + */ + + /* function exit code */ + __pyx_L0:; + __Pyx_XGIVEREF(__pyx_r); + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1034 + * # Versions of the import_* functions which are more suitable for + * # Cython code. + * cdef inline int import_array() except -1: # <<<<<<<<<<<<<< + * try: + * _import_array() + */ + +static CYTHON_INLINE int __pyx_f_5numpy_import_array(void) { + int __pyx_r; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + PyObject *__pyx_t_2 = NULL; + PyObject *__pyx_t_3 = NULL; + int __pyx_t_4; + PyObject *__pyx_t_5 = NULL; + PyObject *__pyx_t_6 = NULL; + PyObject *__pyx_t_7 = NULL; + PyObject *__pyx_t_8 = NULL; + __Pyx_RefNannySetupContext("import_array", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1035 + * # Cython code. + * cdef inline int import_array() except -1: + * try: # <<<<<<<<<<<<<< + * _import_array() + * except Exception: + */ + { + __Pyx_PyThreadState_declare + __Pyx_PyThreadState_assign + __Pyx_ExceptionSave(&__pyx_t_1, &__pyx_t_2, &__pyx_t_3); + __Pyx_XGOTREF(__pyx_t_1); + __Pyx_XGOTREF(__pyx_t_2); + __Pyx_XGOTREF(__pyx_t_3); + /*try:*/ { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1036 + * cdef inline int import_array() except -1: + * try: + * _import_array() # <<<<<<<<<<<<<< + * except Exception: + * raise ImportError("numpy.core.multiarray failed to import") + */ + __pyx_t_4 = _import_array(); if (unlikely(__pyx_t_4 == ((int)-1))) __PYX_ERR(1, 1036, __pyx_L3_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1035 + * # Cython code. + * cdef inline int import_array() except -1: + * try: # <<<<<<<<<<<<<< + * _import_array() + * except Exception: + */ + } + __Pyx_XDECREF(__pyx_t_1); __pyx_t_1 = 0; + __Pyx_XDECREF(__pyx_t_2); __pyx_t_2 = 0; + __Pyx_XDECREF(__pyx_t_3); __pyx_t_3 = 0; + goto __pyx_L8_try_end; + __pyx_L3_error:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1037 + * try: + * _import_array() + * except Exception: # <<<<<<<<<<<<<< + * raise ImportError("numpy.core.multiarray failed to import") + * + */ + __pyx_t_4 = __Pyx_PyErr_ExceptionMatches(((PyObject *)(&((PyTypeObject*)PyExc_Exception)[0]))); + if (__pyx_t_4) { + __Pyx_AddTraceback("numpy.import_array", __pyx_clineno, __pyx_lineno, __pyx_filename); + if (__Pyx_GetException(&__pyx_t_5, &__pyx_t_6, &__pyx_t_7) < 0) __PYX_ERR(1, 1037, __pyx_L5_except_error) + __Pyx_GOTREF(__pyx_t_5); + __Pyx_GOTREF(__pyx_t_6); + __Pyx_GOTREF(__pyx_t_7); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1038 + * _import_array() + * except Exception: + * raise ImportError("numpy.core.multiarray failed to import") # <<<<<<<<<<<<<< + * + * cdef inline int import_umath() except -1: + */ + __pyx_t_8 = __Pyx_PyObject_Call(__pyx_builtin_ImportError, __pyx_tuple__6, NULL); if (unlikely(!__pyx_t_8)) __PYX_ERR(1, 1038, __pyx_L5_except_error) + __Pyx_GOTREF(__pyx_t_8); + __Pyx_Raise(__pyx_t_8, 0, 0, 0); + __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; + __PYX_ERR(1, 1038, __pyx_L5_except_error) + } + goto __pyx_L5_except_error; + __pyx_L5_except_error:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1035 + * # Cython code. + * cdef inline int import_array() except -1: + * try: # <<<<<<<<<<<<<< + * _import_array() + * except Exception: + */ + __Pyx_XGIVEREF(__pyx_t_1); + __Pyx_XGIVEREF(__pyx_t_2); + __Pyx_XGIVEREF(__pyx_t_3); + __Pyx_ExceptionReset(__pyx_t_1, __pyx_t_2, __pyx_t_3); + goto __pyx_L1_error; + __pyx_L8_try_end:; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1034 + * # Versions of the import_* functions which are more suitable for + * # Cython code. + * cdef inline int import_array() except -1: # <<<<<<<<<<<<<< + * try: + * _import_array() + */ + + /* function exit code */ + __pyx_r = 0; + goto __pyx_L0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_5); + __Pyx_XDECREF(__pyx_t_6); + __Pyx_XDECREF(__pyx_t_7); + __Pyx_XDECREF(__pyx_t_8); + __Pyx_AddTraceback("numpy.import_array", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = -1; + __pyx_L0:; + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1040 + * raise ImportError("numpy.core.multiarray failed to import") + * + * cdef inline int import_umath() except -1: # <<<<<<<<<<<<<< + * try: + * _import_umath() + */ + +static CYTHON_INLINE int __pyx_f_5numpy_import_umath(void) { + int __pyx_r; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + PyObject *__pyx_t_2 = NULL; + PyObject *__pyx_t_3 = NULL; + int __pyx_t_4; + PyObject *__pyx_t_5 = NULL; + PyObject *__pyx_t_6 = NULL; + PyObject *__pyx_t_7 = NULL; + PyObject *__pyx_t_8 = NULL; + __Pyx_RefNannySetupContext("import_umath", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1041 + * + * cdef inline int import_umath() except -1: + * try: # <<<<<<<<<<<<<< + * _import_umath() + * except Exception: + */ + { + __Pyx_PyThreadState_declare + __Pyx_PyThreadState_assign + __Pyx_ExceptionSave(&__pyx_t_1, &__pyx_t_2, &__pyx_t_3); + __Pyx_XGOTREF(__pyx_t_1); + __Pyx_XGOTREF(__pyx_t_2); + __Pyx_XGOTREF(__pyx_t_3); + /*try:*/ { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1042 + * cdef inline int import_umath() except -1: + * try: + * _import_umath() # <<<<<<<<<<<<<< + * except Exception: + * raise ImportError("numpy.core.umath failed to import") + */ + __pyx_t_4 = _import_umath(); if (unlikely(__pyx_t_4 == ((int)-1))) __PYX_ERR(1, 1042, __pyx_L3_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1041 + * + * cdef inline int import_umath() except -1: + * try: # <<<<<<<<<<<<<< + * _import_umath() + * except Exception: + */ + } + __Pyx_XDECREF(__pyx_t_1); __pyx_t_1 = 0; + __Pyx_XDECREF(__pyx_t_2); __pyx_t_2 = 0; + __Pyx_XDECREF(__pyx_t_3); __pyx_t_3 = 0; + goto __pyx_L8_try_end; + __pyx_L3_error:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1043 + * try: + * _import_umath() + * except Exception: # <<<<<<<<<<<<<< + * raise ImportError("numpy.core.umath failed to import") + * + */ + __pyx_t_4 = __Pyx_PyErr_ExceptionMatches(((PyObject *)(&((PyTypeObject*)PyExc_Exception)[0]))); + if (__pyx_t_4) { + __Pyx_AddTraceback("numpy.import_umath", __pyx_clineno, __pyx_lineno, __pyx_filename); + if (__Pyx_GetException(&__pyx_t_5, &__pyx_t_6, &__pyx_t_7) < 0) __PYX_ERR(1, 1043, __pyx_L5_except_error) + __Pyx_GOTREF(__pyx_t_5); + __Pyx_GOTREF(__pyx_t_6); + __Pyx_GOTREF(__pyx_t_7); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1044 + * _import_umath() + * except Exception: + * raise ImportError("numpy.core.umath failed to import") # <<<<<<<<<<<<<< + * + * cdef inline int import_ufunc() except -1: + */ + __pyx_t_8 = __Pyx_PyObject_Call(__pyx_builtin_ImportError, __pyx_tuple__7, NULL); if (unlikely(!__pyx_t_8)) __PYX_ERR(1, 1044, __pyx_L5_except_error) + __Pyx_GOTREF(__pyx_t_8); + __Pyx_Raise(__pyx_t_8, 0, 0, 0); + __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; + __PYX_ERR(1, 1044, __pyx_L5_except_error) + } + goto __pyx_L5_except_error; + __pyx_L5_except_error:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1041 + * + * cdef inline int import_umath() except -1: + * try: # <<<<<<<<<<<<<< + * _import_umath() + * except Exception: + */ + __Pyx_XGIVEREF(__pyx_t_1); + __Pyx_XGIVEREF(__pyx_t_2); + __Pyx_XGIVEREF(__pyx_t_3); + __Pyx_ExceptionReset(__pyx_t_1, __pyx_t_2, __pyx_t_3); + goto __pyx_L1_error; + __pyx_L8_try_end:; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1040 + * raise ImportError("numpy.core.multiarray failed to import") + * + * cdef inline int import_umath() except -1: # <<<<<<<<<<<<<< + * try: + * _import_umath() + */ + + /* function exit code */ + __pyx_r = 0; + goto __pyx_L0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_5); + __Pyx_XDECREF(__pyx_t_6); + __Pyx_XDECREF(__pyx_t_7); + __Pyx_XDECREF(__pyx_t_8); + __Pyx_AddTraceback("numpy.import_umath", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = -1; + __pyx_L0:; + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +/* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1046 + * raise ImportError("numpy.core.umath failed to import") + * + * cdef inline int import_ufunc() except -1: # <<<<<<<<<<<<<< + * try: + * _import_umath() + */ + +static CYTHON_INLINE int __pyx_f_5numpy_import_ufunc(void) { + int __pyx_r; + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + PyObject *__pyx_t_2 = NULL; + PyObject *__pyx_t_3 = NULL; + int __pyx_t_4; + PyObject *__pyx_t_5 = NULL; + PyObject *__pyx_t_6 = NULL; + PyObject *__pyx_t_7 = NULL; + PyObject *__pyx_t_8 = NULL; + __Pyx_RefNannySetupContext("import_ufunc", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1047 + * + * cdef inline int import_ufunc() except -1: + * try: # <<<<<<<<<<<<<< + * _import_umath() + * except Exception: + */ + { + __Pyx_PyThreadState_declare + __Pyx_PyThreadState_assign + __Pyx_ExceptionSave(&__pyx_t_1, &__pyx_t_2, &__pyx_t_3); + __Pyx_XGOTREF(__pyx_t_1); + __Pyx_XGOTREF(__pyx_t_2); + __Pyx_XGOTREF(__pyx_t_3); + /*try:*/ { + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1048 + * cdef inline int import_ufunc() except -1: + * try: + * _import_umath() # <<<<<<<<<<<<<< + * except Exception: + * raise ImportError("numpy.core.umath failed to import") + */ + __pyx_t_4 = _import_umath(); if (unlikely(__pyx_t_4 == ((int)-1))) __PYX_ERR(1, 1048, __pyx_L3_error) + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1047 + * + * cdef inline int import_ufunc() except -1: + * try: # <<<<<<<<<<<<<< + * _import_umath() + * except Exception: + */ + } + __Pyx_XDECREF(__pyx_t_1); __pyx_t_1 = 0; + __Pyx_XDECREF(__pyx_t_2); __pyx_t_2 = 0; + __Pyx_XDECREF(__pyx_t_3); __pyx_t_3 = 0; + goto __pyx_L8_try_end; + __pyx_L3_error:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1049 + * try: + * _import_umath() + * except Exception: # <<<<<<<<<<<<<< + * raise ImportError("numpy.core.umath failed to import") + */ + __pyx_t_4 = __Pyx_PyErr_ExceptionMatches(((PyObject *)(&((PyTypeObject*)PyExc_Exception)[0]))); + if (__pyx_t_4) { + __Pyx_AddTraceback("numpy.import_ufunc", __pyx_clineno, __pyx_lineno, __pyx_filename); + if (__Pyx_GetException(&__pyx_t_5, &__pyx_t_6, &__pyx_t_7) < 0) __PYX_ERR(1, 1049, __pyx_L5_except_error) + __Pyx_GOTREF(__pyx_t_5); + __Pyx_GOTREF(__pyx_t_6); + __Pyx_GOTREF(__pyx_t_7); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1050 + * _import_umath() + * except Exception: + * raise ImportError("numpy.core.umath failed to import") # <<<<<<<<<<<<<< + */ + __pyx_t_8 = __Pyx_PyObject_Call(__pyx_builtin_ImportError, __pyx_tuple__7, NULL); if (unlikely(!__pyx_t_8)) __PYX_ERR(1, 1050, __pyx_L5_except_error) + __Pyx_GOTREF(__pyx_t_8); + __Pyx_Raise(__pyx_t_8, 0, 0, 0); + __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; + __PYX_ERR(1, 1050, __pyx_L5_except_error) + } + goto __pyx_L5_except_error; + __pyx_L5_except_error:; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1047 + * + * cdef inline int import_ufunc() except -1: + * try: # <<<<<<<<<<<<<< + * _import_umath() + * except Exception: + */ + __Pyx_XGIVEREF(__pyx_t_1); + __Pyx_XGIVEREF(__pyx_t_2); + __Pyx_XGIVEREF(__pyx_t_3); + __Pyx_ExceptionReset(__pyx_t_1, __pyx_t_2, __pyx_t_3); + goto __pyx_L1_error; + __pyx_L8_try_end:; + } + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1046 + * raise ImportError("numpy.core.umath failed to import") + * + * cdef inline int import_ufunc() except -1: # <<<<<<<<<<<<<< + * try: + * _import_umath() + */ + + /* function exit code */ + __pyx_r = 0; + goto __pyx_L0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_5); + __Pyx_XDECREF(__pyx_t_6); + __Pyx_XDECREF(__pyx_t_7); + __Pyx_XDECREF(__pyx_t_8); + __Pyx_AddTraceback("numpy.import_ufunc", __pyx_clineno, __pyx_lineno, __pyx_filename); + __pyx_r = -1; + __pyx_L0:; + __Pyx_RefNannyFinishContext(); + return __pyx_r; +} + +static PyMethodDef __pyx_methods[] = { + {0, 0, 0, 0} +}; + +#if PY_MAJOR_VERSION >= 3 +#if CYTHON_PEP489_MULTI_PHASE_INIT +static PyObject* __pyx_pymod_create(PyObject *spec, PyModuleDef *def); /*proto*/ +static int __pyx_pymod_exec_pooling_inner(PyObject* module); /*proto*/ +static PyModuleDef_Slot __pyx_moduledef_slots[] = { + {Py_mod_create, (void*)__pyx_pymod_create}, + {Py_mod_exec, (void*)__pyx_pymod_exec_pooling_inner}, + {0, NULL} +}; +#endif + +static struct PyModuleDef __pyx_moduledef = { + PyModuleDef_HEAD_INIT, + "pooling_inner", + __pyx_k_Optimized_cython_functions_for_c, /* m_doc */ + #if CYTHON_PEP489_MULTI_PHASE_INIT + 0, /* m_size */ + #else + -1, /* m_size */ + #endif + __pyx_methods /* m_methods */, + #if CYTHON_PEP489_MULTI_PHASE_INIT + __pyx_moduledef_slots, /* m_slots */ + #else + NULL, /* m_reload */ + #endif + NULL, /* m_traverse */ + NULL, /* m_clear */ + NULL /* m_free */ +}; +#endif +#ifndef CYTHON_SMALL_CODE +#if defined(__clang__) + #define CYTHON_SMALL_CODE +#elif defined(__GNUC__) && (__GNUC__ > 4 || (__GNUC__ == 4 && __GNUC_MINOR__ >= 3)) + #define CYTHON_SMALL_CODE __attribute__((cold)) +#else + #define CYTHON_SMALL_CODE +#endif +#endif + +static __Pyx_StringTabEntry __pyx_string_tab[] = { + {&__pyx_n_s_FAST_VERSION, __pyx_k_FAST_VERSION, sizeof(__pyx_k_FAST_VERSION), 0, 0, 1, 1}, + {&__pyx_kp_u_Format_string_allocated_too_shor, __pyx_k_Format_string_allocated_too_shor, sizeof(__pyx_k_Format_string_allocated_too_shor), 0, 1, 0, 0}, + {&__pyx_kp_u_Format_string_allocated_too_shor_2, __pyx_k_Format_string_allocated_too_shor_2, sizeof(__pyx_k_Format_string_allocated_too_shor_2), 0, 1, 0, 0}, + {&__pyx_n_s_ImportError, __pyx_k_ImportError, sizeof(__pyx_k_ImportError), 0, 0, 1, 1}, + {&__pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_k_MAX_NGRAMS_IN_BATCH, sizeof(__pyx_k_MAX_NGRAMS_IN_BATCH), 0, 0, 1, 1}, + {&__pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_k_MAX_WORDS_IN_BATCH, sizeof(__pyx_k_MAX_WORDS_IN_BATCH), 0, 0, 1, 1}, + {&__pyx_kp_u_Non_native_byte_order_not_suppor, __pyx_k_Non_native_byte_order_not_suppor, sizeof(__pyx_k_Non_native_byte_order_not_suppor), 0, 1, 0, 0}, + {&__pyx_n_s_RuntimeError, __pyx_k_RuntimeError, sizeof(__pyx_k_RuntimeError), 0, 0, 1, 1}, + {&__pyx_n_s_ValueError, __pyx_k_ValueError, sizeof(__pyx_k_ValueError), 0, 0, 1, 1}, + {&__pyx_n_s__8, __pyx_k__8, sizeof(__pyx_k__8), 0, 0, 1, 1}, + {&__pyx_n_s_cline_in_traceback, __pyx_k_cline_in_traceback, sizeof(__pyx_k_cline_in_traceback), 0, 0, 1, 1}, + {&__pyx_n_s_eff_sentences, __pyx_k_eff_sentences, sizeof(__pyx_k_eff_sentences), 0, 0, 1, 1}, + {&__pyx_n_s_eff_words, __pyx_k_eff_words, sizeof(__pyx_k_eff_words), 0, 0, 1, 1}, + {&__pyx_n_s_fblas, __pyx_k_fblas, sizeof(__pyx_k_fblas), 0, 0, 1, 1}, + {&__pyx_n_s_ft, __pyx_k_ft, sizeof(__pyx_k_ft), 0, 0, 1, 1}, + {&__pyx_n_s_import, __pyx_k_import, sizeof(__pyx_k_import), 0, 0, 1, 1}, + {&__pyx_n_s_indexed_sentences, __pyx_k_indexed_sentences, sizeof(__pyx_k_indexed_sentences), 0, 0, 1, 1}, + {&__pyx_n_s_init, __pyx_k_init, sizeof(__pyx_k_init), 0, 0, 1, 1}, + {&__pyx_n_s_is_ft, __pyx_k_is_ft, sizeof(__pyx_k_is_ft), 0, 0, 1, 1}, + {&__pyx_n_s_main, __pyx_k_main, sizeof(__pyx_k_main), 0, 0, 1, 1}, + {&__pyx_n_s_memory, __pyx_k_memory, sizeof(__pyx_k_memory), 0, 0, 1, 1}, + {&__pyx_n_s_model, __pyx_k_model, sizeof(__pyx_k_model), 0, 0, 1, 1}, + {&__pyx_n_s_name, __pyx_k_name, sizeof(__pyx_k_name), 0, 0, 1, 1}, + {&__pyx_kp_u_ndarray_is_not_C_contiguous, __pyx_k_ndarray_is_not_C_contiguous, sizeof(__pyx_k_ndarray_is_not_C_contiguous), 0, 1, 0, 0}, + {&__pyx_kp_u_ndarray_is_not_Fortran_contiguou, __pyx_k_ndarray_is_not_Fortran_contiguou, sizeof(__pyx_k_ndarray_is_not_Fortran_contiguou), 0, 1, 0, 0}, + {&__pyx_n_s_np, __pyx_k_np, sizeof(__pyx_k_np), 0, 0, 1, 1}, + {&__pyx_n_s_numpy, __pyx_k_numpy, sizeof(__pyx_k_numpy), 0, 0, 1, 1}, + {&__pyx_kp_s_numpy_core_multiarray_failed_to, __pyx_k_numpy_core_multiarray_failed_to, sizeof(__pyx_k_numpy_core_multiarray_failed_to), 0, 0, 1, 0}, + {&__pyx_kp_s_numpy_core_umath_failed_to_impor, __pyx_k_numpy_core_umath_failed_to_impor, sizeof(__pyx_k_numpy_core_umath_failed_to_impor), 0, 0, 1, 0}, + {&__pyx_n_s_pooling_inner, __pyx_k_pooling_inner, sizeof(__pyx_k_pooling_inner), 0, 0, 1, 1}, + {&__pyx_kp_s_pooling_inner_pyx, __pyx_k_pooling_inner_pyx, sizeof(__pyx_k_pooling_inner_pyx), 0, 0, 1, 0}, + {&__pyx_n_s_range, __pyx_k_range, sizeof(__pyx_k_range), 0, 0, 1, 1}, + {&__pyx_n_s_scipy_linalg_blas, __pyx_k_scipy_linalg_blas, sizeof(__pyx_k_scipy_linalg_blas), 0, 0, 1, 1}, + {&__pyx_n_s_target, __pyx_k_target, sizeof(__pyx_k_target), 0, 0, 1, 1}, + {&__pyx_n_s_test, __pyx_k_test, sizeof(__pyx_k_test), 0, 0, 1, 1}, + {&__pyx_n_s_train_pooling_cy, __pyx_k_train_pooling_cy, sizeof(__pyx_k_train_pooling_cy), 0, 0, 1, 1}, + {&__pyx_kp_u_unknown_dtype_code_in_numpy_pxd, __pyx_k_unknown_dtype_code_in_numpy_pxd, sizeof(__pyx_k_unknown_dtype_code_in_numpy_pxd), 0, 1, 0, 0}, + {&__pyx_n_s_vocab, __pyx_k_vocab, sizeof(__pyx_k_vocab), 0, 0, 1, 1}, + {&__pyx_n_s_w2v, __pyx_k_w2v, sizeof(__pyx_k_w2v), 0, 0, 1, 1}, + {&__pyx_n_s_wv, __pyx_k_wv, sizeof(__pyx_k_wv), 0, 0, 1, 1}, + {0, 0, 0, 0, 0, 0, 0} +}; +static CYTHON_SMALL_CODE int __Pyx_InitCachedBuiltins(void) { + __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 84, __pyx_L1_error) + __pyx_builtin_ValueError = __Pyx_GetBuiltinName(__pyx_n_s_ValueError); if (!__pyx_builtin_ValueError) __PYX_ERR(1, 272, __pyx_L1_error) + __pyx_builtin_RuntimeError = __Pyx_GetBuiltinName(__pyx_n_s_RuntimeError); if (!__pyx_builtin_RuntimeError) __PYX_ERR(1, 856, __pyx_L1_error) + __pyx_builtin_ImportError = __Pyx_GetBuiltinName(__pyx_n_s_ImportError); if (!__pyx_builtin_ImportError) __PYX_ERR(1, 1038, __pyx_L1_error) + return 0; + __pyx_L1_error:; + return -1; +} + +static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__Pyx_InitCachedConstants", 0); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":272 + * if ((flags & pybuf.PyBUF_C_CONTIGUOUS == pybuf.PyBUF_C_CONTIGUOUS) + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_C_CONTIGUOUS)): + * raise ValueError(u"ndarray is not C contiguous") # <<<<<<<<<<<<<< + * + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) + */ + __pyx_tuple_ = PyTuple_Pack(1, __pyx_kp_u_ndarray_is_not_C_contiguous); if (unlikely(!__pyx_tuple_)) __PYX_ERR(1, 272, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple_); + __Pyx_GIVEREF(__pyx_tuple_); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":276 + * if ((flags & pybuf.PyBUF_F_CONTIGUOUS == pybuf.PyBUF_F_CONTIGUOUS) + * and not PyArray_CHKFLAGS(self, NPY_ARRAY_F_CONTIGUOUS)): + * raise ValueError(u"ndarray is not Fortran contiguous") # <<<<<<<<<<<<<< + * + * info.buf = PyArray_DATA(self) + */ + __pyx_tuple__2 = PyTuple_Pack(1, __pyx_kp_u_ndarray_is_not_Fortran_contiguou); if (unlikely(!__pyx_tuple__2)) __PYX_ERR(1, 276, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__2); + __Pyx_GIVEREF(__pyx_tuple__2); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":306 + * if ((descr.byteorder == c'>' and little_endian) or + * (descr.byteorder == c'<' and not little_endian)): + * raise ValueError(u"Non-native byte order not supported") # <<<<<<<<<<<<<< + * if t == NPY_BYTE: f = "b" + * elif t == NPY_UBYTE: f = "B" + */ + __pyx_tuple__3 = PyTuple_Pack(1, __pyx_kp_u_Non_native_byte_order_not_suppor); if (unlikely(!__pyx_tuple__3)) __PYX_ERR(1, 306, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__3); + __Pyx_GIVEREF(__pyx_tuple__3); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":856 + * + * if (end - f) - (new_offset - offset[0]) < 15: + * raise RuntimeError(u"Format string allocated too short, see comment in numpy.pxd") # <<<<<<<<<<<<<< + * + * if ((child.byteorder == c'>' and little_endian) or + */ + __pyx_tuple__4 = PyTuple_Pack(1, __pyx_kp_u_Format_string_allocated_too_shor); if (unlikely(!__pyx_tuple__4)) __PYX_ERR(1, 856, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__4); + __Pyx_GIVEREF(__pyx_tuple__4); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":880 + * t = child.type_num + * if end - f < 5: + * raise RuntimeError(u"Format string allocated too short.") # <<<<<<<<<<<<<< + * + * # Until ticket #99 is fixed, use integers to avoid warnings + */ + __pyx_tuple__5 = PyTuple_Pack(1, __pyx_kp_u_Format_string_allocated_too_shor_2); if (unlikely(!__pyx_tuple__5)) __PYX_ERR(1, 880, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__5); + __Pyx_GIVEREF(__pyx_tuple__5); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1038 + * _import_array() + * except Exception: + * raise ImportError("numpy.core.multiarray failed to import") # <<<<<<<<<<<<<< + * + * cdef inline int import_umath() except -1: + */ + __pyx_tuple__6 = PyTuple_Pack(1, __pyx_kp_s_numpy_core_multiarray_failed_to); if (unlikely(!__pyx_tuple__6)) __PYX_ERR(1, 1038, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__6); + __Pyx_GIVEREF(__pyx_tuple__6); + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1044 + * _import_umath() + * except Exception: + * raise ImportError("numpy.core.umath failed to import") # <<<<<<<<<<<<<< + * + * cdef inline int import_ufunc() except -1: + */ + __pyx_tuple__7 = PyTuple_Pack(1, __pyx_kp_s_numpy_core_umath_failed_to_impor); if (unlikely(!__pyx_tuple__7)) __PYX_ERR(1, 1044, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__7); + __Pyx_GIVEREF(__pyx_tuple__7); + + /* "pooling_inner.pyx":160 + * # saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + * + * def train_pooling_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< + * """Training on a sequence of sentences and update the target ndarray. + * + */ + __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 160, __pyx_L1_error) + __Pyx_GOTREF(__pyx_tuple__9); + __Pyx_GIVEREF(__pyx_tuple__9); + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 160, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 160, __pyx_L1_error) + + /* "pooling_inner.pyx":206 + * return eff_sentences, eff_words + * + * def init(): # <<<<<<<<<<<<<< + * return 1 + * + */ + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 206, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 206, __pyx_L1_error) + __Pyx_RefNannyFinishContext(); + return 0; + __pyx_L1_error:; + __Pyx_RefNannyFinishContext(); + return -1; +} + +static CYTHON_SMALL_CODE int __Pyx_InitGlobals(void) { + if (__Pyx_InitStrings(__pyx_string_tab) < 0) __PYX_ERR(0, 1, __pyx_L1_error); + __pyx_int_1 = PyInt_FromLong(1); if (unlikely(!__pyx_int_1)) __PYX_ERR(0, 1, __pyx_L1_error) + __pyx_int_40 = PyInt_FromLong(40); if (unlikely(!__pyx_int_40)) __PYX_ERR(0, 1, __pyx_L1_error) + __pyx_int_10000 = PyInt_FromLong(10000L); if (unlikely(!__pyx_int_10000)) __PYX_ERR(0, 1, __pyx_L1_error) + return 0; + __pyx_L1_error:; + return -1; +} + +static CYTHON_SMALL_CODE int __Pyx_modinit_global_init_code(void); /*proto*/ +static CYTHON_SMALL_CODE int __Pyx_modinit_variable_export_code(void); /*proto*/ +static CYTHON_SMALL_CODE int __Pyx_modinit_function_export_code(void); /*proto*/ +static CYTHON_SMALL_CODE int __Pyx_modinit_type_init_code(void); /*proto*/ +static CYTHON_SMALL_CODE int __Pyx_modinit_type_import_code(void); /*proto*/ +static CYTHON_SMALL_CODE int __Pyx_modinit_variable_import_code(void); /*proto*/ +static CYTHON_SMALL_CODE int __Pyx_modinit_function_import_code(void); /*proto*/ + +static int __Pyx_modinit_global_init_code(void) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__Pyx_modinit_global_init_code", 0); + /*--- Global init code ---*/ + __Pyx_RefNannyFinishContext(); + return 0; +} + +static int __Pyx_modinit_variable_export_code(void) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__Pyx_modinit_variable_export_code", 0); + /*--- Variable export code ---*/ + __Pyx_RefNannyFinishContext(); + return 0; +} + +static int __Pyx_modinit_function_export_code(void) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__Pyx_modinit_function_export_code", 0); + /*--- Function export code ---*/ + __Pyx_RefNannyFinishContext(); + return 0; +} + +static int __Pyx_modinit_type_init_code(void) { + __Pyx_RefNannyDeclarations + __Pyx_RefNannySetupContext("__Pyx_modinit_type_init_code", 0); + /*--- Type init code ---*/ + __Pyx_RefNannyFinishContext(); + return 0; +} + +static int __Pyx_modinit_type_import_code(void) { + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("__Pyx_modinit_type_import_code", 0); + /*--- Type import code ---*/ + __pyx_t_1 = PyImport_ImportModule(__Pyx_BUILTIN_MODULE_NAME); if (unlikely(!__pyx_t_1)) __PYX_ERR(2, 9, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_ptype_7cpython_4type_type = __Pyx_ImportType(__pyx_t_1, __Pyx_BUILTIN_MODULE_NAME, "type", + #if defined(PYPY_VERSION_NUM) && PYPY_VERSION_NUM < 0x050B0000 + sizeof(PyTypeObject), + #else + sizeof(PyHeapTypeObject), + #endif + __Pyx_ImportType_CheckSize_Warn); + if (!__pyx_ptype_7cpython_4type_type) __PYX_ERR(2, 9, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_1 = PyImport_ImportModule("numpy"); if (unlikely(!__pyx_t_1)) __PYX_ERR(1, 206, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_ptype_5numpy_dtype = __Pyx_ImportType(__pyx_t_1, "numpy", "dtype", sizeof(PyArray_Descr), __Pyx_ImportType_CheckSize_Ignore); + if (!__pyx_ptype_5numpy_dtype) __PYX_ERR(1, 206, __pyx_L1_error) + __pyx_ptype_5numpy_flatiter = __Pyx_ImportType(__pyx_t_1, "numpy", "flatiter", sizeof(PyArrayIterObject), __Pyx_ImportType_CheckSize_Warn); + if (!__pyx_ptype_5numpy_flatiter) __PYX_ERR(1, 229, __pyx_L1_error) + __pyx_ptype_5numpy_broadcast = __Pyx_ImportType(__pyx_t_1, "numpy", "broadcast", sizeof(PyArrayMultiIterObject), __Pyx_ImportType_CheckSize_Warn); + if (!__pyx_ptype_5numpy_broadcast) __PYX_ERR(1, 233, __pyx_L1_error) + __pyx_ptype_5numpy_ndarray = __Pyx_ImportType(__pyx_t_1, "numpy", "ndarray", sizeof(PyArrayObject), __Pyx_ImportType_CheckSize_Ignore); + if (!__pyx_ptype_5numpy_ndarray) __PYX_ERR(1, 242, __pyx_L1_error) + __pyx_ptype_5numpy_ufunc = __Pyx_ImportType(__pyx_t_1, "numpy", "ufunc", sizeof(PyUFuncObject), __Pyx_ImportType_CheckSize_Warn); + if (!__pyx_ptype_5numpy_ufunc) __PYX_ERR(1, 918, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __Pyx_RefNannyFinishContext(); + return 0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_RefNannyFinishContext(); + return -1; +} + +static int __Pyx_modinit_variable_import_code(void) { + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("__Pyx_modinit_variable_import_code", 0); + /*--- Variable import code ---*/ + __pyx_t_1 = PyImport_ImportModule("average_inner"); if (!__pyx_t_1) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportVoidPtr(__pyx_t_1, "saxpy", (void **)&__pyx_vp_13average_inner_saxpy, "__pyx_t_13average_inner_saxpy_ptr") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportVoidPtr(__pyx_t_1, "sscal", (void **)&__pyx_vp_13average_inner_sscal, "__pyx_t_13average_inner_sscal_ptr") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportVoidPtr(__pyx_t_1, "ONE", (void **)&__pyx_vp_13average_inner_ONE, "int") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportVoidPtr(__pyx_t_1, "ZERO", (void **)&__pyx_vp_13average_inner_ZERO, "int") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportVoidPtr(__pyx_t_1, "ONEF", (void **)&__pyx_vp_13average_inner_ONEF, "__pyx_t_13average_inner_REAL_t") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportVoidPtr(__pyx_t_1, "ZEROF", (void **)&__pyx_vp_13average_inner_ZEROF, "__pyx_t_13average_inner_REAL_t") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + Py_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __Pyx_RefNannyFinishContext(); + return 0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_RefNannyFinishContext(); + return -1; +} + +static int __Pyx_modinit_function_import_code(void) { + __Pyx_RefNannyDeclarations + PyObject *__pyx_t_1 = NULL; + __Pyx_RefNannySetupContext("__Pyx_modinit_function_import_code", 0); + /*--- Function import code ---*/ + __pyx_t_1 = PyImport_ImportModule("average_inner"); if (!__pyx_t_1) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportFunction(__pyx_t_1, "init_base_s2v_config", (void (**)(void))&__pyx_f_13average_inner_init_base_s2v_config, "PyObject *(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportFunction(__pyx_t_1, "init_ft_s2v_config", (void (**)(void))&__pyx_f_13average_inner_init_ft_s2v_config, "PyObject *(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportFunction(__pyx_t_1, "populate_base_s2v_config", (void (**)(void))&__pyx_f_13average_inner_populate_base_s2v_config, "PyObject *(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportFunction(__pyx_t_1, "populate_ft_s2v_config", (void (**)(void))&__pyx_f_13average_inner_populate_ft_s2v_config, "PyObject *(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + Py_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __Pyx_RefNannyFinishContext(); + return 0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_RefNannyFinishContext(); + return -1; +} + + +#if PY_MAJOR_VERSION < 3 +#ifdef CYTHON_NO_PYINIT_EXPORT +#define __Pyx_PyMODINIT_FUNC void +#else +#define __Pyx_PyMODINIT_FUNC PyMODINIT_FUNC +#endif +#else +#ifdef CYTHON_NO_PYINIT_EXPORT +#define __Pyx_PyMODINIT_FUNC PyObject * +#else +#define __Pyx_PyMODINIT_FUNC PyMODINIT_FUNC +#endif +#endif + + +#if PY_MAJOR_VERSION < 3 +__Pyx_PyMODINIT_FUNC initpooling_inner(void) CYTHON_SMALL_CODE; /*proto*/ +__Pyx_PyMODINIT_FUNC initpooling_inner(void) +#else +__Pyx_PyMODINIT_FUNC PyInit_pooling_inner(void) CYTHON_SMALL_CODE; /*proto*/ +__Pyx_PyMODINIT_FUNC PyInit_pooling_inner(void) +#if CYTHON_PEP489_MULTI_PHASE_INIT +{ + return PyModuleDef_Init(&__pyx_moduledef); +} +static CYTHON_SMALL_CODE int __Pyx_check_single_interpreter(void) { + #if PY_VERSION_HEX >= 0x030700A1 + static PY_INT64_T main_interpreter_id = -1; + PY_INT64_T current_id = PyInterpreterState_GetID(PyThreadState_Get()->interp); + if (main_interpreter_id == -1) { + main_interpreter_id = current_id; + return (unlikely(current_id == -1)) ? -1 : 0; + } else if (unlikely(main_interpreter_id != current_id)) + #else + static PyInterpreterState *main_interpreter = NULL; + PyInterpreterState *current_interpreter = PyThreadState_Get()->interp; + if (!main_interpreter) { + main_interpreter = current_interpreter; + } else if (unlikely(main_interpreter != current_interpreter)) + #endif + { + PyErr_SetString( + PyExc_ImportError, + "Interpreter change detected - this module can only be loaded into one interpreter per process."); + return -1; + } + return 0; +} +static CYTHON_SMALL_CODE int __Pyx_copy_spec_to_module(PyObject *spec, PyObject *moddict, const char* from_name, const char* to_name, int allow_none) { + PyObject *value = PyObject_GetAttrString(spec, from_name); + int result = 0; + if (likely(value)) { + if (allow_none || value != Py_None) { + result = PyDict_SetItemString(moddict, to_name, value); + } + Py_DECREF(value); + } else if (PyErr_ExceptionMatches(PyExc_AttributeError)) { + PyErr_Clear(); + } else { + result = -1; + } + return result; +} +static CYTHON_SMALL_CODE PyObject* __pyx_pymod_create(PyObject *spec, CYTHON_UNUSED PyModuleDef *def) { + PyObject *module = NULL, *moddict, *modname; + if (__Pyx_check_single_interpreter()) + return NULL; + if (__pyx_m) + return __Pyx_NewRef(__pyx_m); + modname = PyObject_GetAttrString(spec, "name"); + if (unlikely(!modname)) goto bad; + module = PyModule_NewObject(modname); + Py_DECREF(modname); + if (unlikely(!module)) goto bad; + moddict = PyModule_GetDict(module); + if (unlikely(!moddict)) goto bad; + if (unlikely(__Pyx_copy_spec_to_module(spec, moddict, "loader", "__loader__", 1) < 0)) goto bad; + if (unlikely(__Pyx_copy_spec_to_module(spec, moddict, "origin", "__file__", 1) < 0)) goto bad; + if (unlikely(__Pyx_copy_spec_to_module(spec, moddict, "parent", "__package__", 1) < 0)) goto bad; + if (unlikely(__Pyx_copy_spec_to_module(spec, moddict, "submodule_search_locations", "__path__", 0) < 0)) goto bad; + return module; +bad: + Py_XDECREF(module); + return NULL; +} + + +static CYTHON_SMALL_CODE int __pyx_pymod_exec_pooling_inner(PyObject *__pyx_pyinit_module) +#endif +#endif +{ + PyObject *__pyx_t_1 = NULL; + PyObject *__pyx_t_2 = NULL; + __Pyx_RefNannyDeclarations + #if CYTHON_PEP489_MULTI_PHASE_INIT + if (__pyx_m) { + if (__pyx_m == __pyx_pyinit_module) return 0; + PyErr_SetString(PyExc_RuntimeError, "Module 'pooling_inner' has already been imported. Re-initialisation is not supported."); + return -1; + } + #elif PY_MAJOR_VERSION >= 3 + if (__pyx_m) return __Pyx_NewRef(__pyx_m); + #endif + #if CYTHON_REFNANNY +__Pyx_RefNanny = __Pyx_RefNannyImportAPI("refnanny"); +if (!__Pyx_RefNanny) { + PyErr_Clear(); + __Pyx_RefNanny = __Pyx_RefNannyImportAPI("Cython.Runtime.refnanny"); + if (!__Pyx_RefNanny) + Py_FatalError("failed to import 'refnanny' module"); +} +#endif + __Pyx_RefNannySetupContext("__Pyx_PyMODINIT_FUNC PyInit_pooling_inner(void)", 0); + if (__Pyx_check_binary_version() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #ifdef __Pxy_PyFrame_Initialize_Offsets + __Pxy_PyFrame_Initialize_Offsets(); + #endif + __pyx_empty_tuple = PyTuple_New(0); if (unlikely(!__pyx_empty_tuple)) __PYX_ERR(0, 1, __pyx_L1_error) + __pyx_empty_bytes = PyBytes_FromStringAndSize("", 0); if (unlikely(!__pyx_empty_bytes)) __PYX_ERR(0, 1, __pyx_L1_error) + __pyx_empty_unicode = PyUnicode_FromStringAndSize("", 0); if (unlikely(!__pyx_empty_unicode)) __PYX_ERR(0, 1, __pyx_L1_error) + #ifdef __Pyx_CyFunction_USED + if (__pyx_CyFunction_init() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + #ifdef __Pyx_FusedFunction_USED + if (__pyx_FusedFunction_init() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + #ifdef __Pyx_Coroutine_USED + if (__pyx_Coroutine_init() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + #ifdef __Pyx_Generator_USED + if (__pyx_Generator_init() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + #ifdef __Pyx_AsyncGen_USED + if (__pyx_AsyncGen_init() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + #ifdef __Pyx_StopAsyncIteration_USED + if (__pyx_StopAsyncIteration_init() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + /*--- Library function declarations ---*/ + /*--- Threads initialization code ---*/ + #if defined(__PYX_FORCE_INIT_THREADS) && __PYX_FORCE_INIT_THREADS + #ifdef WITH_THREAD /* Python build with threading support? */ + PyEval_InitThreads(); + #endif + #endif + /*--- Module creation code ---*/ + #if CYTHON_PEP489_MULTI_PHASE_INIT + __pyx_m = __pyx_pyinit_module; + Py_INCREF(__pyx_m); + #else + #if PY_MAJOR_VERSION < 3 + __pyx_m = Py_InitModule4("pooling_inner", __pyx_methods, __pyx_k_Optimized_cython_functions_for_c, 0, PYTHON_API_VERSION); Py_XINCREF(__pyx_m); + #else + __pyx_m = PyModule_Create(&__pyx_moduledef); + #endif + if (unlikely(!__pyx_m)) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + __pyx_d = PyModule_GetDict(__pyx_m); if (unlikely(!__pyx_d)) __PYX_ERR(0, 1, __pyx_L1_error) + Py_INCREF(__pyx_d); + __pyx_b = PyImport_AddModule(__Pyx_BUILTIN_MODULE_NAME); if (unlikely(!__pyx_b)) __PYX_ERR(0, 1, __pyx_L1_error) + Py_INCREF(__pyx_b); + __pyx_cython_runtime = PyImport_AddModule((char *) "cython_runtime"); if (unlikely(!__pyx_cython_runtime)) __PYX_ERR(0, 1, __pyx_L1_error) + Py_INCREF(__pyx_cython_runtime); + if (PyObject_SetAttrString(__pyx_m, "__builtins__", __pyx_b) < 0) __PYX_ERR(0, 1, __pyx_L1_error); + /*--- Initialize various global constants etc. ---*/ + if (__Pyx_InitGlobals() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #if PY_MAJOR_VERSION < 3 && (__PYX_DEFAULT_STRING_ENCODING_IS_ASCII || __PYX_DEFAULT_STRING_ENCODING_IS_DEFAULT) + if (__Pyx_init_sys_getdefaultencoding_params() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + if (__pyx_module_is_main_pooling_inner) { + if (PyObject_SetAttr(__pyx_m, __pyx_n_s_name, __pyx_n_s_main) < 0) __PYX_ERR(0, 1, __pyx_L1_error) + } + #if PY_MAJOR_VERSION >= 3 + { + PyObject *modules = PyImport_GetModuleDict(); if (unlikely(!modules)) __PYX_ERR(0, 1, __pyx_L1_error) + if (!PyDict_GetItemString(modules, "pooling_inner")) { + if (unlikely(PyDict_SetItemString(modules, "pooling_inner", __pyx_m) < 0)) __PYX_ERR(0, 1, __pyx_L1_error) + } + } + #endif + /*--- Builtin init code ---*/ + if (__Pyx_InitCachedBuiltins() < 0) goto __pyx_L1_error; + /*--- Constants init code ---*/ + if (__Pyx_InitCachedConstants() < 0) goto __pyx_L1_error; + /*--- Global type/function init code ---*/ + (void)__Pyx_modinit_global_init_code(); + (void)__Pyx_modinit_variable_export_code(); + (void)__Pyx_modinit_function_export_code(); + (void)__Pyx_modinit_type_init_code(); + if (unlikely(__Pyx_modinit_type_import_code() != 0)) goto __pyx_L1_error; + if (unlikely(__Pyx_modinit_variable_import_code() != 0)) goto __pyx_L1_error; + if (unlikely(__Pyx_modinit_function_import_code() != 0)) goto __pyx_L1_error; + /*--- Execution code ---*/ + #if defined(__Pyx_Generator_USED) || defined(__Pyx_Coroutine_USED) + if (__Pyx_patch_abc() < 0) __PYX_ERR(0, 1, __pyx_L1_error) + #endif + + /* "pooling_inner.pyx":14 + * + * import cython + * import numpy as np # <<<<<<<<<<<<<< + * + * cimport numpy as np + */ + __pyx_t_1 = __Pyx_Import(__pyx_n_s_numpy, 0, -1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 14, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + if (PyDict_SetItem(__pyx_d, __pyx_n_s_np, __pyx_t_1) < 0) __PYX_ERR(0, 14, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "pooling_inner.pyx":20 + * from libc.string cimport memset + * + * import scipy.linalg.blas as fblas # <<<<<<<<<<<<<< + * + * from average_inner cimport ( + */ + __pyx_t_1 = PyList_New(1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 20, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_INCREF(__pyx_n_s__8); + __Pyx_GIVEREF(__pyx_n_s__8); + PyList_SET_ITEM(__pyx_t_1, 0, __pyx_n_s__8); + __pyx_t_2 = __Pyx_Import(__pyx_n_s_scipy_linalg_blas, __pyx_t_1, -1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 20, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_2); + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_2) < 0) __PYX_ERR(0, 20, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; + + /* "pooling_inner.pyx":160 + * # saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + * + * def train_pooling_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< + * """Training on a sequence of sentences and update the target ndarray. + * + */ + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 160, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_2); + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 160, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; + + /* "pooling_inner.pyx":206 + * return eff_sentences, eff_words + * + * def init(): # <<<<<<<<<<<<<< + * return 1 + * + */ + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 206, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_2); + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 206, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; + + /* "pooling_inner.pyx":209 + * return 1 + * + * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< + * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS + * FAST_VERSION = init() + */ + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 209, __pyx_L1_error) + + /* "pooling_inner.pyx":210 + * + * MAX_WORDS_IN_BATCH = MAX_WORDS + * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< + * FAST_VERSION = init() + */ + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 210, __pyx_L1_error) + + /* "pooling_inner.pyx":211 + * MAX_WORDS_IN_BATCH = MAX_WORDS + * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS + * FAST_VERSION = init() # <<<<<<<<<<<<<< + */ + __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 211, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_2); + __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 211, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 211, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "pooling_inner.pyx":1 + * #!/usr/bin/env cython # <<<<<<<<<<<<<< + * # cython: boundscheck=False + * # cython: wraparound=False + */ + __pyx_t_1 = __Pyx_PyDict_NewPresized(0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 1, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + if (PyDict_SetItem(__pyx_d, __pyx_n_s_test, __pyx_t_1) < 0) __PYX_ERR(0, 1, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + + /* "../../../../../../../../anaconda3/envs/fsedev/lib/python3.7/site-packages/Cython/Includes/numpy/__init__.pxd":1046 + * raise ImportError("numpy.core.umath failed to import") + * + * cdef inline int import_ufunc() except -1: # <<<<<<<<<<<<<< + * try: + * _import_umath() + */ + + /*--- Wrapped vars code ---*/ + + goto __pyx_L0; + __pyx_L1_error:; + __Pyx_XDECREF(__pyx_t_1); + __Pyx_XDECREF(__pyx_t_2); + if (__pyx_m) { + if (__pyx_d) { + __Pyx_AddTraceback("init pooling_inner", __pyx_clineno, __pyx_lineno, __pyx_filename); + } + Py_CLEAR(__pyx_m); + } else if (!PyErr_Occurred()) { + PyErr_SetString(PyExc_ImportError, "init pooling_inner"); + } + __pyx_L0:; + __Pyx_RefNannyFinishContext(); + #if CYTHON_PEP489_MULTI_PHASE_INIT + return (__pyx_m != NULL) ? 0 : -1; + #elif PY_MAJOR_VERSION >= 3 + return __pyx_m; + #else + return; + #endif +} + +/* --- Runtime support code --- */ +/* Refnanny */ +#if CYTHON_REFNANNY +static __Pyx_RefNannyAPIStruct *__Pyx_RefNannyImportAPI(const char *modname) { + PyObject *m = NULL, *p = NULL; + void *r = NULL; + m = PyImport_ImportModule(modname); + if (!m) goto end; + p = PyObject_GetAttrString(m, "RefNannyAPI"); + if (!p) goto end; + r = PyLong_AsVoidPtr(p); +end: + Py_XDECREF(p); + Py_XDECREF(m); + return (__Pyx_RefNannyAPIStruct *)r; +} +#endif + +/* PyObjectGetAttrStr */ +#if CYTHON_USE_TYPE_SLOTS +static CYTHON_INLINE PyObject* __Pyx_PyObject_GetAttrStr(PyObject* obj, PyObject* attr_name) { + PyTypeObject* tp = Py_TYPE(obj); + if (likely(tp->tp_getattro)) + return tp->tp_getattro(obj, attr_name); +#if PY_MAJOR_VERSION < 3 + if (likely(tp->tp_getattr)) + return tp->tp_getattr(obj, PyString_AS_STRING(attr_name)); +#endif + return PyObject_GetAttr(obj, attr_name); +} +#endif + +/* GetBuiltinName */ +static PyObject *__Pyx_GetBuiltinName(PyObject *name) { + PyObject* result = __Pyx_PyObject_GetAttrStr(__pyx_b, name); + if (unlikely(!result)) { + PyErr_Format(PyExc_NameError, +#if PY_MAJOR_VERSION >= 3 + "name '%U' is not defined", name); +#else + "name '%.200s' is not defined", PyString_AS_STRING(name)); +#endif + } + return result; +} + +/* RaiseArgTupleInvalid */ +static void __Pyx_RaiseArgtupleInvalid( + const char* func_name, + int exact, + Py_ssize_t num_min, + Py_ssize_t num_max, + Py_ssize_t num_found) +{ + Py_ssize_t num_expected; + const char *more_or_less; + if (num_found < num_min) { + num_expected = num_min; + more_or_less = "at least"; + } else { + num_expected = num_max; + more_or_less = "at most"; + } + if (exact) { + more_or_less = "exactly"; + } + PyErr_Format(PyExc_TypeError, + "%.200s() takes %.8s %" CYTHON_FORMAT_SSIZE_T "d positional argument%.1s (%" CYTHON_FORMAT_SSIZE_T "d given)", + func_name, more_or_less, num_expected, + (num_expected == 1) ? "" : "s", num_found); +} + +/* RaiseDoubleKeywords */ +static void __Pyx_RaiseDoubleKeywordsError( + const char* func_name, + PyObject* kw_name) +{ + PyErr_Format(PyExc_TypeError, + #if PY_MAJOR_VERSION >= 3 + "%s() got multiple values for keyword argument '%U'", func_name, kw_name); + #else + "%s() got multiple values for keyword argument '%s'", func_name, + PyString_AsString(kw_name)); + #endif +} + +/* ParseKeywords */ +static int __Pyx_ParseOptionalKeywords( + PyObject *kwds, + PyObject **argnames[], + PyObject *kwds2, + PyObject *values[], + Py_ssize_t num_pos_args, + const char* function_name) +{ + PyObject *key = 0, *value = 0; + Py_ssize_t pos = 0; + PyObject*** name; + PyObject*** first_kw_arg = argnames + num_pos_args; + while (PyDict_Next(kwds, &pos, &key, &value)) { + name = first_kw_arg; + while (*name && (**name != key)) name++; + if (*name) { + values[name-argnames] = value; + continue; + } + name = first_kw_arg; + #if PY_MAJOR_VERSION < 3 + if (likely(PyString_CheckExact(key)) || likely(PyString_Check(key))) { + while (*name) { + if ((CYTHON_COMPILING_IN_PYPY || PyString_GET_SIZE(**name) == PyString_GET_SIZE(key)) + && _PyString_Eq(**name, key)) { + values[name-argnames] = value; + break; + } + name++; + } + if (*name) continue; + else { + PyObject*** argname = argnames; + while (argname != first_kw_arg) { + if ((**argname == key) || ( + (CYTHON_COMPILING_IN_PYPY || PyString_GET_SIZE(**argname) == PyString_GET_SIZE(key)) + && _PyString_Eq(**argname, key))) { + goto arg_passed_twice; + } + argname++; + } + } + } else + #endif + if (likely(PyUnicode_Check(key))) { + while (*name) { + int cmp = (**name == key) ? 0 : + #if !CYTHON_COMPILING_IN_PYPY && PY_MAJOR_VERSION >= 3 + (PyUnicode_GET_SIZE(**name) != PyUnicode_GET_SIZE(key)) ? 1 : + #endif + PyUnicode_Compare(**name, key); + if (cmp < 0 && unlikely(PyErr_Occurred())) goto bad; + if (cmp == 0) { + values[name-argnames] = value; + break; + } + name++; + } + if (*name) continue; + else { + PyObject*** argname = argnames; + while (argname != first_kw_arg) { + int cmp = (**argname == key) ? 0 : + #if !CYTHON_COMPILING_IN_PYPY && PY_MAJOR_VERSION >= 3 + (PyUnicode_GET_SIZE(**argname) != PyUnicode_GET_SIZE(key)) ? 1 : + #endif + PyUnicode_Compare(**argname, key); + if (cmp < 0 && unlikely(PyErr_Occurred())) goto bad; + if (cmp == 0) goto arg_passed_twice; + argname++; + } + } + } else + goto invalid_keyword_type; + if (kwds2) { + if (unlikely(PyDict_SetItem(kwds2, key, value))) goto bad; + } else { + goto invalid_keyword; + } + } + return 0; +arg_passed_twice: + __Pyx_RaiseDoubleKeywordsError(function_name, key); + goto bad; +invalid_keyword_type: + PyErr_Format(PyExc_TypeError, + "%.200s() keywords must be strings", function_name); + goto bad; +invalid_keyword: + PyErr_Format(PyExc_TypeError, + #if PY_MAJOR_VERSION < 3 + "%.200s() got an unexpected keyword argument '%.200s'", + function_name, PyString_AsString(key)); + #else + "%s() got an unexpected keyword argument '%U'", + function_name, key); + #endif +bad: + return -1; +} + +/* RaiseTooManyValuesToUnpack */ +static CYTHON_INLINE void __Pyx_RaiseTooManyValuesError(Py_ssize_t expected) { + PyErr_Format(PyExc_ValueError, + "too many values to unpack (expected %" CYTHON_FORMAT_SSIZE_T "d)", expected); +} + +/* RaiseNeedMoreValuesToUnpack */ +static CYTHON_INLINE void __Pyx_RaiseNeedMoreValuesError(Py_ssize_t index) { + PyErr_Format(PyExc_ValueError, + "need more than %" CYTHON_FORMAT_SSIZE_T "d value%.1s to unpack", + index, (index == 1) ? "" : "s"); +} + +/* IterFinish */ +static CYTHON_INLINE int __Pyx_IterFinish(void) { +#if CYTHON_FAST_THREAD_STATE + PyThreadState *tstate = __Pyx_PyThreadState_Current; + PyObject* exc_type = tstate->curexc_type; + if (unlikely(exc_type)) { + if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) { + PyObject *exc_value, *exc_tb; + exc_value = tstate->curexc_value; + exc_tb = tstate->curexc_traceback; + tstate->curexc_type = 0; + tstate->curexc_value = 0; + tstate->curexc_traceback = 0; + Py_DECREF(exc_type); + Py_XDECREF(exc_value); + Py_XDECREF(exc_tb); + return 0; + } else { + return -1; + } + } + return 0; +#else + if (unlikely(PyErr_Occurred())) { + if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) { + PyErr_Clear(); + return 0; + } else { + return -1; + } + } + return 0; +#endif +} + +/* UnpackItemEndCheck */ +static int __Pyx_IternextUnpackEndCheck(PyObject *retval, Py_ssize_t expected) { + if (unlikely(retval)) { + Py_DECREF(retval); + __Pyx_RaiseTooManyValuesError(expected); + return -1; + } else { + return __Pyx_IterFinish(); + } + return 0; +} + +/* PyObjectCall */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE PyObject* __Pyx_PyObject_Call(PyObject *func, PyObject *arg, PyObject *kw) { + PyObject *result; + ternaryfunc call = func->ob_type->tp_call; + if (unlikely(!call)) + return PyObject_Call(func, arg, kw); + if (unlikely(Py_EnterRecursiveCall((char*)" while calling a Python object"))) + return NULL; + result = (*call)(func, arg, kw); + Py_LeaveRecursiveCall(); + if (unlikely(!result) && unlikely(!PyErr_Occurred())) { + PyErr_SetString( + PyExc_SystemError, + "NULL result without error in PyObject_Call"); + } + return result; +} +#endif + +/* PyErrFetchRestore */ +#if CYTHON_FAST_THREAD_STATE +static CYTHON_INLINE void __Pyx_ErrRestoreInState(PyThreadState *tstate, PyObject *type, PyObject *value, PyObject *tb) { + PyObject *tmp_type, *tmp_value, *tmp_tb; + tmp_type = tstate->curexc_type; + tmp_value = tstate->curexc_value; + tmp_tb = tstate->curexc_traceback; + tstate->curexc_type = type; + tstate->curexc_value = value; + tstate->curexc_traceback = tb; + Py_XDECREF(tmp_type); + Py_XDECREF(tmp_value); + Py_XDECREF(tmp_tb); +} +static CYTHON_INLINE void __Pyx_ErrFetchInState(PyThreadState *tstate, PyObject **type, PyObject **value, PyObject **tb) { + *type = tstate->curexc_type; + *value = tstate->curexc_value; + *tb = tstate->curexc_traceback; + tstate->curexc_type = 0; + tstate->curexc_value = 0; + tstate->curexc_traceback = 0; +} +#endif + +/* RaiseException */ +#if PY_MAJOR_VERSION < 3 +static void __Pyx_Raise(PyObject *type, PyObject *value, PyObject *tb, + CYTHON_UNUSED PyObject *cause) { + __Pyx_PyThreadState_declare + Py_XINCREF(type); + if (!value || value == Py_None) + value = NULL; + else + Py_INCREF(value); + if (!tb || tb == Py_None) + tb = NULL; + else { + Py_INCREF(tb); + if (!PyTraceBack_Check(tb)) { + PyErr_SetString(PyExc_TypeError, + "raise: arg 3 must be a traceback or None"); + goto raise_error; + } + } + if (PyType_Check(type)) { +#if CYTHON_COMPILING_IN_PYPY + if (!value) { + Py_INCREF(Py_None); + value = Py_None; + } +#endif + PyErr_NormalizeException(&type, &value, &tb); + } else { + if (value) { + PyErr_SetString(PyExc_TypeError, + "instance exception may not have a separate value"); + goto raise_error; + } + value = type; + type = (PyObject*) Py_TYPE(type); + Py_INCREF(type); + if (!PyType_IsSubtype((PyTypeObject *)type, (PyTypeObject *)PyExc_BaseException)) { + PyErr_SetString(PyExc_TypeError, + "raise: exception class must be a subclass of BaseException"); + goto raise_error; + } + } + __Pyx_PyThreadState_assign + __Pyx_ErrRestore(type, value, tb); + return; +raise_error: + Py_XDECREF(value); + Py_XDECREF(type); + Py_XDECREF(tb); + return; +} +#else +static void __Pyx_Raise(PyObject *type, PyObject *value, PyObject *tb, PyObject *cause) { + PyObject* owned_instance = NULL; + if (tb == Py_None) { + tb = 0; + } else if (tb && !PyTraceBack_Check(tb)) { + PyErr_SetString(PyExc_TypeError, + "raise: arg 3 must be a traceback or None"); + goto bad; + } + if (value == Py_None) + value = 0; + if (PyExceptionInstance_Check(type)) { + if (value) { + PyErr_SetString(PyExc_TypeError, + "instance exception may not have a separate value"); + goto bad; + } + value = type; + type = (PyObject*) Py_TYPE(value); + } else if (PyExceptionClass_Check(type)) { + PyObject *instance_class = NULL; + if (value && PyExceptionInstance_Check(value)) { + instance_class = (PyObject*) Py_TYPE(value); + if (instance_class != type) { + int is_subclass = PyObject_IsSubclass(instance_class, type); + if (!is_subclass) { + instance_class = NULL; + } else if (unlikely(is_subclass == -1)) { + goto bad; + } else { + type = instance_class; + } + } + } + if (!instance_class) { + PyObject *args; + if (!value) + args = PyTuple_New(0); + else if (PyTuple_Check(value)) { + Py_INCREF(value); + args = value; + } else + args = PyTuple_Pack(1, value); + if (!args) + goto bad; + owned_instance = PyObject_Call(type, args, NULL); + Py_DECREF(args); + if (!owned_instance) + goto bad; + value = owned_instance; + if (!PyExceptionInstance_Check(value)) { + PyErr_Format(PyExc_TypeError, + "calling %R should have returned an instance of " + "BaseException, not %R", + type, Py_TYPE(value)); + goto bad; + } + } + } else { + PyErr_SetString(PyExc_TypeError, + "raise: exception class must be a subclass of BaseException"); + goto bad; + } + if (cause) { + PyObject *fixed_cause; + if (cause == Py_None) { + fixed_cause = NULL; + } else if (PyExceptionClass_Check(cause)) { + fixed_cause = PyObject_CallObject(cause, NULL); + if (fixed_cause == NULL) + goto bad; + } else if (PyExceptionInstance_Check(cause)) { + fixed_cause = cause; + Py_INCREF(fixed_cause); + } else { + PyErr_SetString(PyExc_TypeError, + "exception causes must derive from " + "BaseException"); + goto bad; + } + PyException_SetCause(value, fixed_cause); + } + PyErr_SetObject(type, value); + if (tb) { +#if CYTHON_COMPILING_IN_PYPY + PyObject *tmp_type, *tmp_value, *tmp_tb; + PyErr_Fetch(&tmp_type, &tmp_value, &tmp_tb); + Py_INCREF(tb); + PyErr_Restore(tmp_type, tmp_value, tb); + Py_XDECREF(tmp_tb); +#else + PyThreadState *tstate = __Pyx_PyThreadState_Current; + PyObject* tmp_tb = tstate->curexc_traceback; + if (tb != tmp_tb) { + Py_INCREF(tb); + tstate->curexc_traceback = tb; + Py_XDECREF(tmp_tb); + } +#endif + } +bad: + Py_XDECREF(owned_instance); + return; +} +#endif + +/* PyCFunctionFastCall */ +#if CYTHON_FAST_PYCCALL +static CYTHON_INLINE PyObject * __Pyx_PyCFunction_FastCall(PyObject *func_obj, PyObject **args, Py_ssize_t nargs) { + PyCFunctionObject *func = (PyCFunctionObject*)func_obj; + PyCFunction meth = PyCFunction_GET_FUNCTION(func); + PyObject *self = PyCFunction_GET_SELF(func); + int flags = PyCFunction_GET_FLAGS(func); + assert(PyCFunction_Check(func)); + assert(METH_FASTCALL == (flags & ~(METH_CLASS | METH_STATIC | METH_COEXIST | METH_KEYWORDS | METH_STACKLESS))); + assert(nargs >= 0); + assert(nargs == 0 || args != NULL); + /* _PyCFunction_FastCallDict() must not be called with an exception set, + because it may clear it (directly or indirectly) and so the + caller loses its exception */ + assert(!PyErr_Occurred()); + if ((PY_VERSION_HEX < 0x030700A0) || unlikely(flags & METH_KEYWORDS)) { + return (*((__Pyx_PyCFunctionFastWithKeywords)(void*)meth)) (self, args, nargs, NULL); + } else { + return (*((__Pyx_PyCFunctionFast)(void*)meth)) (self, args, nargs); + } +} +#endif + +/* PyFunctionFastCall */ +#if CYTHON_FAST_PYCALL +static PyObject* __Pyx_PyFunction_FastCallNoKw(PyCodeObject *co, PyObject **args, Py_ssize_t na, + PyObject *globals) { + PyFrameObject *f; + PyThreadState *tstate = __Pyx_PyThreadState_Current; + PyObject **fastlocals; + Py_ssize_t i; + PyObject *result; + assert(globals != NULL); + /* XXX Perhaps we should create a specialized + PyFrame_New() that doesn't take locals, but does + take builtins without sanity checking them. + */ + assert(tstate != NULL); + f = PyFrame_New(tstate, co, globals, NULL); + if (f == NULL) { + return NULL; + } + fastlocals = __Pyx_PyFrame_GetLocalsplus(f); + for (i = 0; i < na; i++) { + Py_INCREF(*args); + fastlocals[i] = *args++; + } + result = PyEval_EvalFrameEx(f,0); + ++tstate->recursion_depth; + Py_DECREF(f); + --tstate->recursion_depth; + return result; +} +#if 1 || PY_VERSION_HEX < 0x030600B1 +static PyObject *__Pyx_PyFunction_FastCallDict(PyObject *func, PyObject **args, Py_ssize_t nargs, PyObject *kwargs) { + PyCodeObject *co = (PyCodeObject *)PyFunction_GET_CODE(func); + PyObject *globals = PyFunction_GET_GLOBALS(func); + PyObject *argdefs = PyFunction_GET_DEFAULTS(func); + PyObject *closure; +#if PY_MAJOR_VERSION >= 3 + PyObject *kwdefs; +#endif + PyObject *kwtuple, **k; + PyObject **d; + Py_ssize_t nd; + Py_ssize_t nk; + PyObject *result; + assert(kwargs == NULL || PyDict_Check(kwargs)); + nk = kwargs ? PyDict_Size(kwargs) : 0; + if (Py_EnterRecursiveCall((char*)" while calling a Python object")) { + return NULL; + } + if ( +#if PY_MAJOR_VERSION >= 3 + co->co_kwonlyargcount == 0 && +#endif + likely(kwargs == NULL || nk == 0) && + co->co_flags == (CO_OPTIMIZED | CO_NEWLOCALS | CO_NOFREE)) { + if (argdefs == NULL && co->co_argcount == nargs) { + result = __Pyx_PyFunction_FastCallNoKw(co, args, nargs, globals); + goto done; + } + else if (nargs == 0 && argdefs != NULL + && co->co_argcount == Py_SIZE(argdefs)) { + /* function called with no arguments, but all parameters have + a default value: use default values as arguments .*/ + args = &PyTuple_GET_ITEM(argdefs, 0); + result =__Pyx_PyFunction_FastCallNoKw(co, args, Py_SIZE(argdefs), globals); + goto done; + } + } + if (kwargs != NULL) { + Py_ssize_t pos, i; + kwtuple = PyTuple_New(2 * nk); + if (kwtuple == NULL) { + result = NULL; + goto done; + } + k = &PyTuple_GET_ITEM(kwtuple, 0); + pos = i = 0; + while (PyDict_Next(kwargs, &pos, &k[i], &k[i+1])) { + Py_INCREF(k[i]); + Py_INCREF(k[i+1]); + i += 2; + } + nk = i / 2; + } + else { + kwtuple = NULL; + k = NULL; + } + closure = PyFunction_GET_CLOSURE(func); +#if PY_MAJOR_VERSION >= 3 + kwdefs = PyFunction_GET_KW_DEFAULTS(func); +#endif + if (argdefs != NULL) { + d = &PyTuple_GET_ITEM(argdefs, 0); + nd = Py_SIZE(argdefs); + } + else { + d = NULL; + nd = 0; + } +#if PY_MAJOR_VERSION >= 3 + result = PyEval_EvalCodeEx((PyObject*)co, globals, (PyObject *)NULL, + args, (int)nargs, + k, (int)nk, + d, (int)nd, kwdefs, closure); +#else + result = PyEval_EvalCodeEx(co, globals, (PyObject *)NULL, + args, (int)nargs, + k, (int)nk, + d, (int)nd, closure); +#endif + Py_XDECREF(kwtuple); +done: + Py_LeaveRecursiveCall(); + return result; +} +#endif +#endif + +/* PyObjectCallMethO */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallMethO(PyObject *func, PyObject *arg) { + PyObject *self, *result; + PyCFunction cfunc; + cfunc = PyCFunction_GET_FUNCTION(func); + self = PyCFunction_GET_SELF(func); + if (unlikely(Py_EnterRecursiveCall((char*)" while calling a Python object"))) + return NULL; + result = cfunc(self, arg); + Py_LeaveRecursiveCall(); + if (unlikely(!result) && unlikely(!PyErr_Occurred())) { + PyErr_SetString( + PyExc_SystemError, + "NULL result without error in PyObject_Call"); + } + return result; +} +#endif + +/* PyObjectCallOneArg */ +#if CYTHON_COMPILING_IN_CPYTHON +static PyObject* __Pyx__PyObject_CallOneArg(PyObject *func, PyObject *arg) { + PyObject *result; + PyObject *args = PyTuple_New(1); + if (unlikely(!args)) return NULL; + Py_INCREF(arg); + PyTuple_SET_ITEM(args, 0, arg); + result = __Pyx_PyObject_Call(func, args, NULL); + Py_DECREF(args); + return result; +} +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallOneArg(PyObject *func, PyObject *arg) { +#if CYTHON_FAST_PYCALL + if (PyFunction_Check(func)) { + return __Pyx_PyFunction_FastCall(func, &arg, 1); + } +#endif + if (likely(PyCFunction_Check(func))) { + if (likely(PyCFunction_GET_FLAGS(func) & METH_O)) { + return __Pyx_PyObject_CallMethO(func, arg); +#if CYTHON_FAST_PYCCALL + } else if (PyCFunction_GET_FLAGS(func) & METH_FASTCALL) { + return __Pyx_PyCFunction_FastCall(func, &arg, 1); +#endif + } + } + return __Pyx__PyObject_CallOneArg(func, arg); +} +#else +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallOneArg(PyObject *func, PyObject *arg) { + PyObject *result; + PyObject *args = PyTuple_Pack(1, arg); + if (unlikely(!args)) return NULL; + result = __Pyx_PyObject_Call(func, args, NULL); + Py_DECREF(args); + return result; +} +#endif + +/* DictGetItem */ +#if PY_MAJOR_VERSION >= 3 && !CYTHON_COMPILING_IN_PYPY +static PyObject *__Pyx_PyDict_GetItem(PyObject *d, PyObject* key) { + PyObject *value; + value = PyDict_GetItemWithError(d, key); + if (unlikely(!value)) { + if (!PyErr_Occurred()) { + if (unlikely(PyTuple_Check(key))) { + PyObject* args = PyTuple_Pack(1, key); + if (likely(args)) { + PyErr_SetObject(PyExc_KeyError, args); + Py_DECREF(args); + } + } else { + PyErr_SetObject(PyExc_KeyError, key); + } + } + return NULL; + } + Py_INCREF(value); + return value; +} +#endif + +/* RaiseNoneIterError */ +static CYTHON_INLINE void __Pyx_RaiseNoneNotIterableError(void) { + PyErr_SetString(PyExc_TypeError, "'NoneType' object is not iterable"); +} + +/* ExtTypeTest */ +static CYTHON_INLINE int __Pyx_TypeTest(PyObject *obj, PyTypeObject *type) { + if (unlikely(!type)) { + PyErr_SetString(PyExc_SystemError, "Missing type object"); + return 0; + } + if (likely(__Pyx_TypeCheck(obj, type))) + return 1; + PyErr_Format(PyExc_TypeError, "Cannot convert %.200s to %.200s", + Py_TYPE(obj)->tp_name, type->tp_name); + return 0; +} + +/* GetTopmostException */ +#if CYTHON_USE_EXC_INFO_STACK +static _PyErr_StackItem * +__Pyx_PyErr_GetTopmostException(PyThreadState *tstate) +{ + _PyErr_StackItem *exc_info = tstate->exc_info; + while ((exc_info->exc_type == NULL || exc_info->exc_type == Py_None) && + exc_info->previous_item != NULL) + { + exc_info = exc_info->previous_item; + } + return exc_info; +} +#endif + +/* SaveResetException */ +#if CYTHON_FAST_THREAD_STATE +static CYTHON_INLINE void __Pyx__ExceptionSave(PyThreadState *tstate, PyObject **type, PyObject **value, PyObject **tb) { + #if CYTHON_USE_EXC_INFO_STACK + _PyErr_StackItem *exc_info = __Pyx_PyErr_GetTopmostException(tstate); + *type = exc_info->exc_type; + *value = exc_info->exc_value; + *tb = exc_info->exc_traceback; + #else + *type = tstate->exc_type; + *value = tstate->exc_value; + *tb = tstate->exc_traceback; + #endif + Py_XINCREF(*type); + Py_XINCREF(*value); + Py_XINCREF(*tb); +} +static CYTHON_INLINE void __Pyx__ExceptionReset(PyThreadState *tstate, PyObject *type, PyObject *value, PyObject *tb) { + PyObject *tmp_type, *tmp_value, *tmp_tb; + #if CYTHON_USE_EXC_INFO_STACK + _PyErr_StackItem *exc_info = tstate->exc_info; + tmp_type = exc_info->exc_type; + tmp_value = exc_info->exc_value; + tmp_tb = exc_info->exc_traceback; + exc_info->exc_type = type; + exc_info->exc_value = value; + exc_info->exc_traceback = tb; + #else + tmp_type = tstate->exc_type; + tmp_value = tstate->exc_value; + tmp_tb = tstate->exc_traceback; + tstate->exc_type = type; + tstate->exc_value = value; + tstate->exc_traceback = tb; + #endif + Py_XDECREF(tmp_type); + Py_XDECREF(tmp_value); + Py_XDECREF(tmp_tb); +} +#endif + +/* PyErrExceptionMatches */ +#if CYTHON_FAST_THREAD_STATE +static int __Pyx_PyErr_ExceptionMatchesTuple(PyObject *exc_type, PyObject *tuple) { + Py_ssize_t i, n; + n = PyTuple_GET_SIZE(tuple); +#if PY_MAJOR_VERSION >= 3 + for (i=0; icurexc_type; + if (exc_type == err) return 1; + if (unlikely(!exc_type)) return 0; + if (unlikely(PyTuple_Check(err))) + return __Pyx_PyErr_ExceptionMatchesTuple(exc_type, err); + return __Pyx_PyErr_GivenExceptionMatches(exc_type, err); +} +#endif + +/* GetException */ +#if CYTHON_FAST_THREAD_STATE +static int __Pyx__GetException(PyThreadState *tstate, PyObject **type, PyObject **value, PyObject **tb) +#else +static int __Pyx_GetException(PyObject **type, PyObject **value, PyObject **tb) +#endif +{ + PyObject *local_type, *local_value, *local_tb; +#if CYTHON_FAST_THREAD_STATE + PyObject *tmp_type, *tmp_value, *tmp_tb; + local_type = tstate->curexc_type; + local_value = tstate->curexc_value; + local_tb = tstate->curexc_traceback; + tstate->curexc_type = 0; + tstate->curexc_value = 0; + tstate->curexc_traceback = 0; +#else + PyErr_Fetch(&local_type, &local_value, &local_tb); +#endif + PyErr_NormalizeException(&local_type, &local_value, &local_tb); +#if CYTHON_FAST_THREAD_STATE + if (unlikely(tstate->curexc_type)) +#else + if (unlikely(PyErr_Occurred())) +#endif + goto bad; + #if PY_MAJOR_VERSION >= 3 + if (local_tb) { + if (unlikely(PyException_SetTraceback(local_value, local_tb) < 0)) + goto bad; + } + #endif + Py_XINCREF(local_tb); + Py_XINCREF(local_type); + Py_XINCREF(local_value); + *type = local_type; + *value = local_value; + *tb = local_tb; +#if CYTHON_FAST_THREAD_STATE + #if CYTHON_USE_EXC_INFO_STACK + { + _PyErr_StackItem *exc_info = tstate->exc_info; + tmp_type = exc_info->exc_type; + tmp_value = exc_info->exc_value; + tmp_tb = exc_info->exc_traceback; + exc_info->exc_type = local_type; + exc_info->exc_value = local_value; + exc_info->exc_traceback = local_tb; + } + #else + tmp_type = tstate->exc_type; + tmp_value = tstate->exc_value; + tmp_tb = tstate->exc_traceback; + tstate->exc_type = local_type; + tstate->exc_value = local_value; + tstate->exc_traceback = local_tb; + #endif + Py_XDECREF(tmp_type); + Py_XDECREF(tmp_value); + Py_XDECREF(tmp_tb); +#else + PyErr_SetExcInfo(local_type, local_value, local_tb); +#endif + return 0; +bad: + *type = 0; + *value = 0; + *tb = 0; + Py_XDECREF(local_type); + Py_XDECREF(local_value); + Py_XDECREF(local_tb); + return -1; +} + +/* TypeImport */ +#ifndef __PYX_HAVE_RT_ImportType +#define __PYX_HAVE_RT_ImportType +static PyTypeObject *__Pyx_ImportType(PyObject *module, const char *module_name, const char *class_name, + size_t size, enum __Pyx_ImportType_CheckSize check_size) +{ + PyObject *result = 0; + char warning[200]; + Py_ssize_t basicsize; +#ifdef Py_LIMITED_API + PyObject *py_basicsize; +#endif + result = PyObject_GetAttrString(module, class_name); + if (!result) + goto bad; + if (!PyType_Check(result)) { + PyErr_Format(PyExc_TypeError, + "%.200s.%.200s is not a type object", + module_name, class_name); + goto bad; + } +#ifndef Py_LIMITED_API + basicsize = ((PyTypeObject *)result)->tp_basicsize; +#else + py_basicsize = PyObject_GetAttrString(result, "__basicsize__"); + if (!py_basicsize) + goto bad; + basicsize = PyLong_AsSsize_t(py_basicsize); + Py_DECREF(py_basicsize); + py_basicsize = 0; + if (basicsize == (Py_ssize_t)-1 && PyErr_Occurred()) + goto bad; +#endif + if ((size_t)basicsize < size) { + PyErr_Format(PyExc_ValueError, + "%.200s.%.200s size changed, may indicate binary incompatibility. " + "Expected %zd from C header, got %zd from PyObject", + module_name, class_name, size, basicsize); + goto bad; + } + if (check_size == __Pyx_ImportType_CheckSize_Error && (size_t)basicsize != size) { + PyErr_Format(PyExc_ValueError, + "%.200s.%.200s size changed, may indicate binary incompatibility. " + "Expected %zd from C header, got %zd from PyObject", + module_name, class_name, size, basicsize); + goto bad; + } + else if (check_size == __Pyx_ImportType_CheckSize_Warn && (size_t)basicsize > size) { + PyOS_snprintf(warning, sizeof(warning), + "%s.%s size changed, may indicate binary incompatibility. " + "Expected %zd from C header, got %zd from PyObject", + module_name, class_name, size, basicsize); + if (PyErr_WarnEx(NULL, warning, 0) < 0) goto bad; + } + return (PyTypeObject *)result; +bad: + Py_XDECREF(result); + return NULL; +} +#endif + +/* Import */ +static PyObject *__Pyx_Import(PyObject *name, PyObject *from_list, int level) { + PyObject *empty_list = 0; + PyObject *module = 0; + PyObject *global_dict = 0; + PyObject *empty_dict = 0; + PyObject *list; + #if PY_MAJOR_VERSION < 3 + PyObject *py_import; + py_import = __Pyx_PyObject_GetAttrStr(__pyx_b, __pyx_n_s_import); + if (!py_import) + goto bad; + #endif + if (from_list) + list = from_list; + else { + empty_list = PyList_New(0); + if (!empty_list) + goto bad; + list = empty_list; + } + global_dict = PyModule_GetDict(__pyx_m); + if (!global_dict) + goto bad; + empty_dict = PyDict_New(); + if (!empty_dict) + goto bad; + { + #if PY_MAJOR_VERSION >= 3 + if (level == -1) { + if (strchr(__Pyx_MODULE_NAME, '.')) { + module = PyImport_ImportModuleLevelObject( + name, global_dict, empty_dict, list, 1); + if (!module) { + if (!PyErr_ExceptionMatches(PyExc_ImportError)) + goto bad; + PyErr_Clear(); + } + } + level = 0; + } + #endif + if (!module) { + #if PY_MAJOR_VERSION < 3 + PyObject *py_level = PyInt_FromLong(level); + if (!py_level) + goto bad; + module = PyObject_CallFunctionObjArgs(py_import, + name, global_dict, empty_dict, list, py_level, (PyObject *)NULL); + Py_DECREF(py_level); + #else + module = PyImport_ImportModuleLevelObject( + name, global_dict, empty_dict, list, level); + #endif + } + } +bad: + #if PY_MAJOR_VERSION < 3 + Py_XDECREF(py_import); + #endif + Py_XDECREF(empty_list); + Py_XDECREF(empty_dict); + return module; +} + +/* PyDictVersioning */ +#if CYTHON_USE_DICT_VERSIONS && CYTHON_USE_TYPE_SLOTS +static CYTHON_INLINE PY_UINT64_T __Pyx_get_tp_dict_version(PyObject *obj) { + PyObject *dict = Py_TYPE(obj)->tp_dict; + return likely(dict) ? __PYX_GET_DICT_VERSION(dict) : 0; +} +static CYTHON_INLINE PY_UINT64_T __Pyx_get_object_dict_version(PyObject *obj) { + PyObject **dictptr = NULL; + Py_ssize_t offset = Py_TYPE(obj)->tp_dictoffset; + if (offset) { +#if CYTHON_COMPILING_IN_CPYTHON + dictptr = (likely(offset > 0)) ? (PyObject **) ((char *)obj + offset) : _PyObject_GetDictPtr(obj); +#else + dictptr = _PyObject_GetDictPtr(obj); +#endif + } + return (dictptr && *dictptr) ? __PYX_GET_DICT_VERSION(*dictptr) : 0; +} +static CYTHON_INLINE int __Pyx_object_dict_version_matches(PyObject* obj, PY_UINT64_T tp_dict_version, PY_UINT64_T obj_dict_version) { + PyObject *dict = Py_TYPE(obj)->tp_dict; + if (unlikely(!dict) || unlikely(tp_dict_version != __PYX_GET_DICT_VERSION(dict))) + return 0; + return obj_dict_version == __Pyx_get_object_dict_version(obj); +} +#endif + +/* GetModuleGlobalName */ +#if CYTHON_USE_DICT_VERSIONS +static PyObject *__Pyx__GetModuleGlobalName(PyObject *name, PY_UINT64_T *dict_version, PyObject **dict_cached_value) +#else +static CYTHON_INLINE PyObject *__Pyx__GetModuleGlobalName(PyObject *name) +#endif +{ + PyObject *result; +#if !CYTHON_AVOID_BORROWED_REFS +#if CYTHON_COMPILING_IN_CPYTHON && PY_VERSION_HEX >= 0x030500A1 + result = _PyDict_GetItem_KnownHash(__pyx_d, name, ((PyASCIIObject *) name)->hash); + __PYX_UPDATE_DICT_CACHE(__pyx_d, result, *dict_cached_value, *dict_version) + if (likely(result)) { + return __Pyx_NewRef(result); + } else if (unlikely(PyErr_Occurred())) { + return NULL; + } +#else + result = PyDict_GetItem(__pyx_d, name); + __PYX_UPDATE_DICT_CACHE(__pyx_d, result, *dict_cached_value, *dict_version) + if (likely(result)) { + return __Pyx_NewRef(result); + } +#endif +#else + result = PyObject_GetItem(__pyx_d, name); + __PYX_UPDATE_DICT_CACHE(__pyx_d, result, *dict_cached_value, *dict_version) + if (likely(result)) { + return __Pyx_NewRef(result); + } + PyErr_Clear(); +#endif + return __Pyx_GetBuiltinName(name); +} + +/* PyObjectCallNoArg */ +#if CYTHON_COMPILING_IN_CPYTHON +static CYTHON_INLINE PyObject* __Pyx_PyObject_CallNoArg(PyObject *func) { +#if CYTHON_FAST_PYCALL + if (PyFunction_Check(func)) { + return __Pyx_PyFunction_FastCall(func, NULL, 0); + } +#endif +#ifdef __Pyx_CyFunction_USED + if (likely(PyCFunction_Check(func) || __Pyx_CyFunction_Check(func))) +#else + if (likely(PyCFunction_Check(func))) +#endif + { + if (likely(PyCFunction_GET_FLAGS(func) & METH_NOARGS)) { + return __Pyx_PyObject_CallMethO(func, NULL); + } + } + return __Pyx_PyObject_Call(func, __pyx_empty_tuple, NULL); +} +#endif + +/* CLineInTraceback */ +#ifndef CYTHON_CLINE_IN_TRACEBACK +static int __Pyx_CLineForTraceback(PyThreadState *tstate, int c_line) { + PyObject *use_cline; + PyObject *ptype, *pvalue, *ptraceback; +#if CYTHON_COMPILING_IN_CPYTHON + PyObject **cython_runtime_dict; +#endif + if (unlikely(!__pyx_cython_runtime)) { + return c_line; + } + __Pyx_ErrFetchInState(tstate, &ptype, &pvalue, &ptraceback); +#if CYTHON_COMPILING_IN_CPYTHON + cython_runtime_dict = _PyObject_GetDictPtr(__pyx_cython_runtime); + if (likely(cython_runtime_dict)) { + __PYX_PY_DICT_LOOKUP_IF_MODIFIED( + use_cline, *cython_runtime_dict, + __Pyx_PyDict_GetItemStr(*cython_runtime_dict, __pyx_n_s_cline_in_traceback)) + } else +#endif + { + PyObject *use_cline_obj = __Pyx_PyObject_GetAttrStr(__pyx_cython_runtime, __pyx_n_s_cline_in_traceback); + if (use_cline_obj) { + use_cline = PyObject_Not(use_cline_obj) ? Py_False : Py_True; + Py_DECREF(use_cline_obj); + } else { + PyErr_Clear(); + use_cline = NULL; + } + } + if (!use_cline) { + c_line = 0; + PyObject_SetAttr(__pyx_cython_runtime, __pyx_n_s_cline_in_traceback, Py_False); + } + else if (use_cline == Py_False || (use_cline != Py_True && PyObject_Not(use_cline) != 0)) { + c_line = 0; + } + __Pyx_ErrRestoreInState(tstate, ptype, pvalue, ptraceback); + return c_line; +} +#endif + +/* CodeObjectCache */ +static int __pyx_bisect_code_objects(__Pyx_CodeObjectCacheEntry* entries, int count, int code_line) { + int start = 0, mid = 0, end = count - 1; + if (end >= 0 && code_line > entries[end].code_line) { + return count; + } + while (start < end) { + mid = start + (end - start) / 2; + if (code_line < entries[mid].code_line) { + end = mid; + } else if (code_line > entries[mid].code_line) { + start = mid + 1; + } else { + return mid; + } + } + if (code_line <= entries[mid].code_line) { + return mid; + } else { + return mid + 1; + } +} +static PyCodeObject *__pyx_find_code_object(int code_line) { + PyCodeObject* code_object; + int pos; + if (unlikely(!code_line) || unlikely(!__pyx_code_cache.entries)) { + return NULL; + } + pos = __pyx_bisect_code_objects(__pyx_code_cache.entries, __pyx_code_cache.count, code_line); + if (unlikely(pos >= __pyx_code_cache.count) || unlikely(__pyx_code_cache.entries[pos].code_line != code_line)) { + return NULL; + } + code_object = __pyx_code_cache.entries[pos].code_object; + Py_INCREF(code_object); + return code_object; +} +static void __pyx_insert_code_object(int code_line, PyCodeObject* code_object) { + int pos, i; + __Pyx_CodeObjectCacheEntry* entries = __pyx_code_cache.entries; + if (unlikely(!code_line)) { + return; + } + if (unlikely(!entries)) { + entries = (__Pyx_CodeObjectCacheEntry*)PyMem_Malloc(64*sizeof(__Pyx_CodeObjectCacheEntry)); + if (likely(entries)) { + __pyx_code_cache.entries = entries; + __pyx_code_cache.max_count = 64; + __pyx_code_cache.count = 1; + entries[0].code_line = code_line; + entries[0].code_object = code_object; + Py_INCREF(code_object); + } + return; + } + pos = __pyx_bisect_code_objects(__pyx_code_cache.entries, __pyx_code_cache.count, code_line); + if ((pos < __pyx_code_cache.count) && unlikely(__pyx_code_cache.entries[pos].code_line == code_line)) { + PyCodeObject* tmp = entries[pos].code_object; + entries[pos].code_object = code_object; + Py_DECREF(tmp); + return; + } + if (__pyx_code_cache.count == __pyx_code_cache.max_count) { + int new_max = __pyx_code_cache.max_count + 64; + entries = (__Pyx_CodeObjectCacheEntry*)PyMem_Realloc( + __pyx_code_cache.entries, (size_t)new_max*sizeof(__Pyx_CodeObjectCacheEntry)); + if (unlikely(!entries)) { + return; + } + __pyx_code_cache.entries = entries; + __pyx_code_cache.max_count = new_max; + } + for (i=__pyx_code_cache.count; i>pos; i--) { + entries[i] = entries[i-1]; + } + entries[pos].code_line = code_line; + entries[pos].code_object = code_object; + __pyx_code_cache.count++; + Py_INCREF(code_object); +} + +/* AddTraceback */ +#include "compile.h" +#include "frameobject.h" +#include "traceback.h" +static PyCodeObject* __Pyx_CreateCodeObjectForTraceback( + const char *funcname, int c_line, + int py_line, const char *filename) { + PyCodeObject *py_code = 0; + PyObject *py_srcfile = 0; + PyObject *py_funcname = 0; + #if PY_MAJOR_VERSION < 3 + py_srcfile = PyString_FromString(filename); + #else + py_srcfile = PyUnicode_FromString(filename); + #endif + if (!py_srcfile) goto bad; + if (c_line) { + #if PY_MAJOR_VERSION < 3 + py_funcname = PyString_FromFormat( "%s (%s:%d)", funcname, __pyx_cfilenm, c_line); + #else + py_funcname = PyUnicode_FromFormat( "%s (%s:%d)", funcname, __pyx_cfilenm, c_line); + #endif + } + else { + #if PY_MAJOR_VERSION < 3 + py_funcname = PyString_FromString(funcname); + #else + py_funcname = PyUnicode_FromString(funcname); + #endif + } + if (!py_funcname) goto bad; + py_code = __Pyx_PyCode_New( + 0, + 0, + 0, + 0, + 0, + __pyx_empty_bytes, /*PyObject *code,*/ + __pyx_empty_tuple, /*PyObject *consts,*/ + __pyx_empty_tuple, /*PyObject *names,*/ + __pyx_empty_tuple, /*PyObject *varnames,*/ + __pyx_empty_tuple, /*PyObject *freevars,*/ + __pyx_empty_tuple, /*PyObject *cellvars,*/ + py_srcfile, /*PyObject *filename,*/ + py_funcname, /*PyObject *name,*/ + py_line, + __pyx_empty_bytes /*PyObject *lnotab*/ + ); + Py_DECREF(py_srcfile); + Py_DECREF(py_funcname); + return py_code; +bad: + Py_XDECREF(py_srcfile); + Py_XDECREF(py_funcname); + return NULL; +} +static void __Pyx_AddTraceback(const char *funcname, int c_line, + int py_line, const char *filename) { + PyCodeObject *py_code = 0; + PyFrameObject *py_frame = 0; + PyThreadState *tstate = __Pyx_PyThreadState_Current; + if (c_line) { + c_line = __Pyx_CLineForTraceback(tstate, c_line); + } + py_code = __pyx_find_code_object(c_line ? -c_line : py_line); + if (!py_code) { + py_code = __Pyx_CreateCodeObjectForTraceback( + funcname, c_line, py_line, filename); + if (!py_code) goto bad; + __pyx_insert_code_object(c_line ? -c_line : py_line, py_code); + } + py_frame = PyFrame_New( + tstate, /*PyThreadState *tstate,*/ + py_code, /*PyCodeObject *code,*/ + __pyx_d, /*PyObject *globals,*/ + 0 /*PyObject *locals*/ + ); + if (!py_frame) goto bad; + __Pyx_PyFrame_SetLineNumber(py_frame, py_line); + PyTraceBack_Here(py_frame); +bad: + Py_XDECREF(py_code); + Py_XDECREF(py_frame); +} + +/* CIntToPy */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_npy_uint32(npy_uint32 value) { + const npy_uint32 neg_one = (npy_uint32) ((npy_uint32) 0 - (npy_uint32) 1), const_zero = (npy_uint32) 0; + const int is_unsigned = neg_one > const_zero; + if (is_unsigned) { + if (sizeof(npy_uint32) < sizeof(long)) { + return PyInt_FromLong((long) value); + } else if (sizeof(npy_uint32) <= sizeof(unsigned long)) { + return PyLong_FromUnsignedLong((unsigned long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(npy_uint32) <= sizeof(unsigned PY_LONG_LONG)) { + return PyLong_FromUnsignedLongLong((unsigned PY_LONG_LONG) value); +#endif + } + } else { + if (sizeof(npy_uint32) <= sizeof(long)) { + return PyInt_FromLong((long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(npy_uint32) <= sizeof(PY_LONG_LONG)) { + return PyLong_FromLongLong((PY_LONG_LONG) value); +#endif + } + } + { + int one = 1; int little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&value; + return _PyLong_FromByteArray(bytes, sizeof(npy_uint32), + little, !is_unsigned); + } +} + +/* CIntFromPyVerify */ +#define __PYX_VERIFY_RETURN_INT(target_type, func_type, func_value)\ + __PYX__VERIFY_RETURN_INT(target_type, func_type, func_value, 0) +#define __PYX_VERIFY_RETURN_INT_EXC(target_type, func_type, func_value)\ + __PYX__VERIFY_RETURN_INT(target_type, func_type, func_value, 1) +#define __PYX__VERIFY_RETURN_INT(target_type, func_type, func_value, exc)\ + {\ + func_type value = func_value;\ + if (sizeof(target_type) < sizeof(func_type)) {\ + if (unlikely(value != (func_type) (target_type) value)) {\ + func_type zero = 0;\ + if (exc && unlikely(value == (func_type)-1 && PyErr_Occurred()))\ + return (target_type) -1;\ + if (is_unsigned && unlikely(value < zero))\ + goto raise_neg_overflow;\ + else\ + goto raise_overflow;\ + }\ + }\ + return (target_type) value;\ + } + +/* Declarations */ +#if CYTHON_CCOMPLEX + #ifdef __cplusplus + static CYTHON_INLINE __pyx_t_float_complex __pyx_t_float_complex_from_parts(float x, float y) { + return ::std::complex< float >(x, y); + } + #else + static CYTHON_INLINE __pyx_t_float_complex __pyx_t_float_complex_from_parts(float x, float y) { + return x + y*(__pyx_t_float_complex)_Complex_I; + } + #endif +#else + static CYTHON_INLINE __pyx_t_float_complex __pyx_t_float_complex_from_parts(float x, float y) { + __pyx_t_float_complex z; + z.real = x; + z.imag = y; + return z; + } +#endif + +/* Arithmetic */ +#if CYTHON_CCOMPLEX +#else + static CYTHON_INLINE int __Pyx_c_eq_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + return (a.real == b.real) && (a.imag == b.imag); + } + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_sum_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + __pyx_t_float_complex z; + z.real = a.real + b.real; + z.imag = a.imag + b.imag; + return z; + } + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_diff_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + __pyx_t_float_complex z; + z.real = a.real - b.real; + z.imag = a.imag - b.imag; + return z; + } + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_prod_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + __pyx_t_float_complex z; + z.real = a.real * b.real - a.imag * b.imag; + z.imag = a.real * b.imag + a.imag * b.real; + return z; + } + #if 1 + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_quot_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + if (b.imag == 0) { + return __pyx_t_float_complex_from_parts(a.real / b.real, a.imag / b.real); + } else if (fabsf(b.real) >= fabsf(b.imag)) { + if (b.real == 0 && b.imag == 0) { + return __pyx_t_float_complex_from_parts(a.real / b.real, a.imag / b.imag); + } else { + float r = b.imag / b.real; + float s = (float)(1.0) / (b.real + b.imag * r); + return __pyx_t_float_complex_from_parts( + (a.real + a.imag * r) * s, (a.imag - a.real * r) * s); + } + } else { + float r = b.real / b.imag; + float s = (float)(1.0) / (b.imag + b.real * r); + return __pyx_t_float_complex_from_parts( + (a.real * r + a.imag) * s, (a.imag * r - a.real) * s); + } + } + #else + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_quot_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + if (b.imag == 0) { + return __pyx_t_float_complex_from_parts(a.real / b.real, a.imag / b.real); + } else { + float denom = b.real * b.real + b.imag * b.imag; + return __pyx_t_float_complex_from_parts( + (a.real * b.real + a.imag * b.imag) / denom, + (a.imag * b.real - a.real * b.imag) / denom); + } + } + #endif + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_neg_float(__pyx_t_float_complex a) { + __pyx_t_float_complex z; + z.real = -a.real; + z.imag = -a.imag; + return z; + } + static CYTHON_INLINE int __Pyx_c_is_zero_float(__pyx_t_float_complex a) { + return (a.real == 0) && (a.imag == 0); + } + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_conj_float(__pyx_t_float_complex a) { + __pyx_t_float_complex z; + z.real = a.real; + z.imag = -a.imag; + return z; + } + #if 1 + static CYTHON_INLINE float __Pyx_c_abs_float(__pyx_t_float_complex z) { + #if !defined(HAVE_HYPOT) || defined(_MSC_VER) + return sqrtf(z.real*z.real + z.imag*z.imag); + #else + return hypotf(z.real, z.imag); + #endif + } + static CYTHON_INLINE __pyx_t_float_complex __Pyx_c_pow_float(__pyx_t_float_complex a, __pyx_t_float_complex b) { + __pyx_t_float_complex z; + float r, lnr, theta, z_r, z_theta; + if (b.imag == 0 && b.real == (int)b.real) { + if (b.real < 0) { + float denom = a.real * a.real + a.imag * a.imag; + a.real = a.real / denom; + a.imag = -a.imag / denom; + b.real = -b.real; + } + switch ((int)b.real) { + case 0: + z.real = 1; + z.imag = 0; + return z; + case 1: + return a; + case 2: + return __Pyx_c_prod_float(a, a); + case 3: + z = __Pyx_c_prod_float(a, a); + return __Pyx_c_prod_float(z, a); + case 4: + z = __Pyx_c_prod_float(a, a); + return __Pyx_c_prod_float(z, z); + } + } + if (a.imag == 0) { + if (a.real == 0) { + return a; + } else if (b.imag == 0) { + z.real = powf(a.real, b.real); + z.imag = 0; + return z; + } else if (a.real > 0) { + r = a.real; + theta = 0; + } else { + r = -a.real; + theta = atan2f(0.0, -1.0); + } + } else { + r = __Pyx_c_abs_float(a); + theta = atan2f(a.imag, a.real); + } + lnr = logf(r); + z_r = expf(lnr * b.real - theta * b.imag); + z_theta = theta * b.real + lnr * b.imag; + z.real = z_r * cosf(z_theta); + z.imag = z_r * sinf(z_theta); + return z; + } + #endif +#endif + +/* Declarations */ +#if CYTHON_CCOMPLEX + #ifdef __cplusplus + static CYTHON_INLINE __pyx_t_double_complex __pyx_t_double_complex_from_parts(double x, double y) { + return ::std::complex< double >(x, y); + } + #else + static CYTHON_INLINE __pyx_t_double_complex __pyx_t_double_complex_from_parts(double x, double y) { + return x + y*(__pyx_t_double_complex)_Complex_I; + } + #endif +#else + static CYTHON_INLINE __pyx_t_double_complex __pyx_t_double_complex_from_parts(double x, double y) { + __pyx_t_double_complex z; + z.real = x; + z.imag = y; + return z; + } +#endif + +/* Arithmetic */ +#if CYTHON_CCOMPLEX +#else + static CYTHON_INLINE int __Pyx_c_eq_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + return (a.real == b.real) && (a.imag == b.imag); + } + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_sum_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + __pyx_t_double_complex z; + z.real = a.real + b.real; + z.imag = a.imag + b.imag; + return z; + } + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_diff_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + __pyx_t_double_complex z; + z.real = a.real - b.real; + z.imag = a.imag - b.imag; + return z; + } + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_prod_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + __pyx_t_double_complex z; + z.real = a.real * b.real - a.imag * b.imag; + z.imag = a.real * b.imag + a.imag * b.real; + return z; + } + #if 1 + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_quot_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + if (b.imag == 0) { + return __pyx_t_double_complex_from_parts(a.real / b.real, a.imag / b.real); + } else if (fabs(b.real) >= fabs(b.imag)) { + if (b.real == 0 && b.imag == 0) { + return __pyx_t_double_complex_from_parts(a.real / b.real, a.imag / b.imag); + } else { + double r = b.imag / b.real; + double s = (double)(1.0) / (b.real + b.imag * r); + return __pyx_t_double_complex_from_parts( + (a.real + a.imag * r) * s, (a.imag - a.real * r) * s); + } + } else { + double r = b.real / b.imag; + double s = (double)(1.0) / (b.imag + b.real * r); + return __pyx_t_double_complex_from_parts( + (a.real * r + a.imag) * s, (a.imag * r - a.real) * s); + } + } + #else + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_quot_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + if (b.imag == 0) { + return __pyx_t_double_complex_from_parts(a.real / b.real, a.imag / b.real); + } else { + double denom = b.real * b.real + b.imag * b.imag; + return __pyx_t_double_complex_from_parts( + (a.real * b.real + a.imag * b.imag) / denom, + (a.imag * b.real - a.real * b.imag) / denom); + } + } + #endif + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_neg_double(__pyx_t_double_complex a) { + __pyx_t_double_complex z; + z.real = -a.real; + z.imag = -a.imag; + return z; + } + static CYTHON_INLINE int __Pyx_c_is_zero_double(__pyx_t_double_complex a) { + return (a.real == 0) && (a.imag == 0); + } + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_conj_double(__pyx_t_double_complex a) { + __pyx_t_double_complex z; + z.real = a.real; + z.imag = -a.imag; + return z; + } + #if 1 + static CYTHON_INLINE double __Pyx_c_abs_double(__pyx_t_double_complex z) { + #if !defined(HAVE_HYPOT) || defined(_MSC_VER) + return sqrt(z.real*z.real + z.imag*z.imag); + #else + return hypot(z.real, z.imag); + #endif + } + static CYTHON_INLINE __pyx_t_double_complex __Pyx_c_pow_double(__pyx_t_double_complex a, __pyx_t_double_complex b) { + __pyx_t_double_complex z; + double r, lnr, theta, z_r, z_theta; + if (b.imag == 0 && b.real == (int)b.real) { + if (b.real < 0) { + double denom = a.real * a.real + a.imag * a.imag; + a.real = a.real / denom; + a.imag = -a.imag / denom; + b.real = -b.real; + } + switch ((int)b.real) { + case 0: + z.real = 1; + z.imag = 0; + return z; + case 1: + return a; + case 2: + return __Pyx_c_prod_double(a, a); + case 3: + z = __Pyx_c_prod_double(a, a); + return __Pyx_c_prod_double(z, a); + case 4: + z = __Pyx_c_prod_double(a, a); + return __Pyx_c_prod_double(z, z); + } + } + if (a.imag == 0) { + if (a.real == 0) { + return a; + } else if (b.imag == 0) { + z.real = pow(a.real, b.real); + z.imag = 0; + return z; + } else if (a.real > 0) { + r = a.real; + theta = 0; + } else { + r = -a.real; + theta = atan2(0.0, -1.0); + } + } else { + r = __Pyx_c_abs_double(a); + theta = atan2(a.imag, a.real); + } + lnr = log(r); + z_r = exp(lnr * b.real - theta * b.imag); + z_theta = theta * b.real + lnr * b.imag; + z.real = z_r * cos(z_theta); + z.imag = z_r * sin(z_theta); + return z; + } + #endif +#endif + +/* CIntToPy */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_int(int value) { + const int neg_one = (int) ((int) 0 - (int) 1), const_zero = (int) 0; + const int is_unsigned = neg_one > const_zero; + if (is_unsigned) { + if (sizeof(int) < sizeof(long)) { + return PyInt_FromLong((long) value); + } else if (sizeof(int) <= sizeof(unsigned long)) { + return PyLong_FromUnsignedLong((unsigned long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(int) <= sizeof(unsigned PY_LONG_LONG)) { + return PyLong_FromUnsignedLongLong((unsigned PY_LONG_LONG) value); +#endif + } + } else { + if (sizeof(int) <= sizeof(long)) { + return PyInt_FromLong((long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(int) <= sizeof(PY_LONG_LONG)) { + return PyLong_FromLongLong((PY_LONG_LONG) value); +#endif + } + } + { + int one = 1; int little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&value; + return _PyLong_FromByteArray(bytes, sizeof(int), + little, !is_unsigned); + } +} + +/* CIntToPy */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_enum__NPY_TYPES(enum NPY_TYPES value) { + const enum NPY_TYPES neg_one = (enum NPY_TYPES) ((enum NPY_TYPES) 0 - (enum NPY_TYPES) 1), const_zero = (enum NPY_TYPES) 0; + const int is_unsigned = neg_one > const_zero; + if (is_unsigned) { + if (sizeof(enum NPY_TYPES) < sizeof(long)) { + return PyInt_FromLong((long) value); + } else if (sizeof(enum NPY_TYPES) <= sizeof(unsigned long)) { + return PyLong_FromUnsignedLong((unsigned long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(enum NPY_TYPES) <= sizeof(unsigned PY_LONG_LONG)) { + return PyLong_FromUnsignedLongLong((unsigned PY_LONG_LONG) value); +#endif + } + } else { + if (sizeof(enum NPY_TYPES) <= sizeof(long)) { + return PyInt_FromLong((long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(enum NPY_TYPES) <= sizeof(PY_LONG_LONG)) { + return PyLong_FromLongLong((PY_LONG_LONG) value); +#endif + } + } + { + int one = 1; int little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&value; + return _PyLong_FromByteArray(bytes, sizeof(enum NPY_TYPES), + little, !is_unsigned); + } +} + +/* CIntFromPy */ +static CYTHON_INLINE npy_uint32 __Pyx_PyInt_As_npy_uint32(PyObject *x) { + const npy_uint32 neg_one = (npy_uint32) ((npy_uint32) 0 - (npy_uint32) 1), const_zero = (npy_uint32) 0; + const int is_unsigned = neg_one > const_zero; +#if PY_MAJOR_VERSION < 3 + if (likely(PyInt_Check(x))) { + if (sizeof(npy_uint32) < sizeof(long)) { + __PYX_VERIFY_RETURN_INT(npy_uint32, long, PyInt_AS_LONG(x)) + } else { + long val = PyInt_AS_LONG(x); + if (is_unsigned && unlikely(val < 0)) { + goto raise_neg_overflow; + } + return (npy_uint32) val; + } + } else +#endif + if (likely(PyLong_Check(x))) { + if (is_unsigned) { +#if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)x)->ob_digit; + switch (Py_SIZE(x)) { + case 0: return (npy_uint32) 0; + case 1: __PYX_VERIFY_RETURN_INT(npy_uint32, digit, digits[0]) + case 2: + if (8 * sizeof(npy_uint32) > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, unsigned long, (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) >= 2 * PyLong_SHIFT) { + return (npy_uint32) (((((npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0])); + } + } + break; + case 3: + if (8 * sizeof(npy_uint32) > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, unsigned long, (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) >= 3 * PyLong_SHIFT) { + return (npy_uint32) (((((((npy_uint32)digits[2]) << PyLong_SHIFT) | (npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0])); + } + } + break; + case 4: + if (8 * sizeof(npy_uint32) > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, unsigned long, (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) >= 4 * PyLong_SHIFT) { + return (npy_uint32) (((((((((npy_uint32)digits[3]) << PyLong_SHIFT) | (npy_uint32)digits[2]) << PyLong_SHIFT) | (npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0])); + } + } + break; + } +#endif +#if CYTHON_COMPILING_IN_CPYTHON + if (unlikely(Py_SIZE(x) < 0)) { + goto raise_neg_overflow; + } +#else + { + int result = PyObject_RichCompareBool(x, Py_False, Py_LT); + if (unlikely(result < 0)) + return (npy_uint32) -1; + if (unlikely(result == 1)) + goto raise_neg_overflow; + } +#endif + if (sizeof(npy_uint32) <= sizeof(unsigned long)) { + __PYX_VERIFY_RETURN_INT_EXC(npy_uint32, unsigned long, PyLong_AsUnsignedLong(x)) +#ifdef HAVE_LONG_LONG + } else if (sizeof(npy_uint32) <= sizeof(unsigned PY_LONG_LONG)) { + __PYX_VERIFY_RETURN_INT_EXC(npy_uint32, unsigned PY_LONG_LONG, PyLong_AsUnsignedLongLong(x)) +#endif + } + } else { +#if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)x)->ob_digit; + switch (Py_SIZE(x)) { + case 0: return (npy_uint32) 0; + case -1: __PYX_VERIFY_RETURN_INT(npy_uint32, sdigit, (sdigit) (-(sdigit)digits[0])) + case 1: __PYX_VERIFY_RETURN_INT(npy_uint32, digit, +digits[0]) + case -2: + if (8 * sizeof(npy_uint32) - 1 > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, long, -(long) (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) - 1 > 2 * PyLong_SHIFT) { + return (npy_uint32) (((npy_uint32)-1)*(((((npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0]))); + } + } + break; + case 2: + if (8 * sizeof(npy_uint32) > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, unsigned long, (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) - 1 > 2 * PyLong_SHIFT) { + return (npy_uint32) ((((((npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0]))); + } + } + break; + case -3: + if (8 * sizeof(npy_uint32) - 1 > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, long, -(long) (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) - 1 > 3 * PyLong_SHIFT) { + return (npy_uint32) (((npy_uint32)-1)*(((((((npy_uint32)digits[2]) << PyLong_SHIFT) | (npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0]))); + } + } + break; + case 3: + if (8 * sizeof(npy_uint32) > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, unsigned long, (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) - 1 > 3 * PyLong_SHIFT) { + return (npy_uint32) ((((((((npy_uint32)digits[2]) << PyLong_SHIFT) | (npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0]))); + } + } + break; + case -4: + if (8 * sizeof(npy_uint32) - 1 > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, long, -(long) (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) - 1 > 4 * PyLong_SHIFT) { + return (npy_uint32) (((npy_uint32)-1)*(((((((((npy_uint32)digits[3]) << PyLong_SHIFT) | (npy_uint32)digits[2]) << PyLong_SHIFT) | (npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0]))); + } + } + break; + case 4: + if (8 * sizeof(npy_uint32) > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(npy_uint32, unsigned long, (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(npy_uint32) - 1 > 4 * PyLong_SHIFT) { + return (npy_uint32) ((((((((((npy_uint32)digits[3]) << PyLong_SHIFT) | (npy_uint32)digits[2]) << PyLong_SHIFT) | (npy_uint32)digits[1]) << PyLong_SHIFT) | (npy_uint32)digits[0]))); + } + } + break; + } +#endif + if (sizeof(npy_uint32) <= sizeof(long)) { + __PYX_VERIFY_RETURN_INT_EXC(npy_uint32, long, PyLong_AsLong(x)) +#ifdef HAVE_LONG_LONG + } else if (sizeof(npy_uint32) <= sizeof(PY_LONG_LONG)) { + __PYX_VERIFY_RETURN_INT_EXC(npy_uint32, PY_LONG_LONG, PyLong_AsLongLong(x)) +#endif + } + } + { +#if CYTHON_COMPILING_IN_PYPY && !defined(_PyLong_AsByteArray) + PyErr_SetString(PyExc_RuntimeError, + "_PyLong_AsByteArray() not available in PyPy, cannot convert large numbers"); +#else + npy_uint32 val; + PyObject *v = __Pyx_PyNumber_IntOrLong(x); + #if PY_MAJOR_VERSION < 3 + if (likely(v) && !PyLong_Check(v)) { + PyObject *tmp = v; + v = PyNumber_Long(tmp); + Py_DECREF(tmp); + } + #endif + if (likely(v)) { + int one = 1; int is_little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&val; + int ret = _PyLong_AsByteArray((PyLongObject *)v, + bytes, sizeof(val), + is_little, !is_unsigned); + Py_DECREF(v); + if (likely(!ret)) + return val; + } +#endif + return (npy_uint32) -1; + } + } else { + npy_uint32 val; + PyObject *tmp = __Pyx_PyNumber_IntOrLong(x); + if (!tmp) return (npy_uint32) -1; + val = __Pyx_PyInt_As_npy_uint32(tmp); + Py_DECREF(tmp); + return val; + } +raise_overflow: + PyErr_SetString(PyExc_OverflowError, + "value too large to convert to npy_uint32"); + return (npy_uint32) -1; +raise_neg_overflow: + PyErr_SetString(PyExc_OverflowError, + "can't convert negative value to npy_uint32"); + return (npy_uint32) -1; +} + +/* CIntFromPy */ +static CYTHON_INLINE int __Pyx_PyInt_As_int(PyObject *x) { + const int neg_one = (int) ((int) 0 - (int) 1), const_zero = (int) 0; + const int is_unsigned = neg_one > const_zero; +#if PY_MAJOR_VERSION < 3 + if (likely(PyInt_Check(x))) { + if (sizeof(int) < sizeof(long)) { + __PYX_VERIFY_RETURN_INT(int, long, PyInt_AS_LONG(x)) + } else { + long val = PyInt_AS_LONG(x); + if (is_unsigned && unlikely(val < 0)) { + goto raise_neg_overflow; + } + return (int) val; + } + } else +#endif + if (likely(PyLong_Check(x))) { + if (is_unsigned) { +#if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)x)->ob_digit; + switch (Py_SIZE(x)) { + case 0: return (int) 0; + case 1: __PYX_VERIFY_RETURN_INT(int, digit, digits[0]) + case 2: + if (8 * sizeof(int) > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, unsigned long, (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) >= 2 * PyLong_SHIFT) { + return (int) (((((int)digits[1]) << PyLong_SHIFT) | (int)digits[0])); + } + } + break; + case 3: + if (8 * sizeof(int) > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, unsigned long, (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) >= 3 * PyLong_SHIFT) { + return (int) (((((((int)digits[2]) << PyLong_SHIFT) | (int)digits[1]) << PyLong_SHIFT) | (int)digits[0])); + } + } + break; + case 4: + if (8 * sizeof(int) > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, unsigned long, (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) >= 4 * PyLong_SHIFT) { + return (int) (((((((((int)digits[3]) << PyLong_SHIFT) | (int)digits[2]) << PyLong_SHIFT) | (int)digits[1]) << PyLong_SHIFT) | (int)digits[0])); + } + } + break; + } +#endif +#if CYTHON_COMPILING_IN_CPYTHON + if (unlikely(Py_SIZE(x) < 0)) { + goto raise_neg_overflow; + } +#else + { + int result = PyObject_RichCompareBool(x, Py_False, Py_LT); + if (unlikely(result < 0)) + return (int) -1; + if (unlikely(result == 1)) + goto raise_neg_overflow; + } +#endif + if (sizeof(int) <= sizeof(unsigned long)) { + __PYX_VERIFY_RETURN_INT_EXC(int, unsigned long, PyLong_AsUnsignedLong(x)) +#ifdef HAVE_LONG_LONG + } else if (sizeof(int) <= sizeof(unsigned PY_LONG_LONG)) { + __PYX_VERIFY_RETURN_INT_EXC(int, unsigned PY_LONG_LONG, PyLong_AsUnsignedLongLong(x)) +#endif + } + } else { +#if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)x)->ob_digit; + switch (Py_SIZE(x)) { + case 0: return (int) 0; + case -1: __PYX_VERIFY_RETURN_INT(int, sdigit, (sdigit) (-(sdigit)digits[0])) + case 1: __PYX_VERIFY_RETURN_INT(int, digit, +digits[0]) + case -2: + if (8 * sizeof(int) - 1 > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, long, -(long) (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) - 1 > 2 * PyLong_SHIFT) { + return (int) (((int)-1)*(((((int)digits[1]) << PyLong_SHIFT) | (int)digits[0]))); + } + } + break; + case 2: + if (8 * sizeof(int) > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, unsigned long, (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) - 1 > 2 * PyLong_SHIFT) { + return (int) ((((((int)digits[1]) << PyLong_SHIFT) | (int)digits[0]))); + } + } + break; + case -3: + if (8 * sizeof(int) - 1 > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, long, -(long) (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) - 1 > 3 * PyLong_SHIFT) { + return (int) (((int)-1)*(((((((int)digits[2]) << PyLong_SHIFT) | (int)digits[1]) << PyLong_SHIFT) | (int)digits[0]))); + } + } + break; + case 3: + if (8 * sizeof(int) > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, unsigned long, (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) - 1 > 3 * PyLong_SHIFT) { + return (int) ((((((((int)digits[2]) << PyLong_SHIFT) | (int)digits[1]) << PyLong_SHIFT) | (int)digits[0]))); + } + } + break; + case -4: + if (8 * sizeof(int) - 1 > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, long, -(long) (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) - 1 > 4 * PyLong_SHIFT) { + return (int) (((int)-1)*(((((((((int)digits[3]) << PyLong_SHIFT) | (int)digits[2]) << PyLong_SHIFT) | (int)digits[1]) << PyLong_SHIFT) | (int)digits[0]))); + } + } + break; + case 4: + if (8 * sizeof(int) > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(int, unsigned long, (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(int) - 1 > 4 * PyLong_SHIFT) { + return (int) ((((((((((int)digits[3]) << PyLong_SHIFT) | (int)digits[2]) << PyLong_SHIFT) | (int)digits[1]) << PyLong_SHIFT) | (int)digits[0]))); + } + } + break; + } +#endif + if (sizeof(int) <= sizeof(long)) { + __PYX_VERIFY_RETURN_INT_EXC(int, long, PyLong_AsLong(x)) +#ifdef HAVE_LONG_LONG + } else if (sizeof(int) <= sizeof(PY_LONG_LONG)) { + __PYX_VERIFY_RETURN_INT_EXC(int, PY_LONG_LONG, PyLong_AsLongLong(x)) +#endif + } + } + { +#if CYTHON_COMPILING_IN_PYPY && !defined(_PyLong_AsByteArray) + PyErr_SetString(PyExc_RuntimeError, + "_PyLong_AsByteArray() not available in PyPy, cannot convert large numbers"); +#else + int val; + PyObject *v = __Pyx_PyNumber_IntOrLong(x); + #if PY_MAJOR_VERSION < 3 + if (likely(v) && !PyLong_Check(v)) { + PyObject *tmp = v; + v = PyNumber_Long(tmp); + Py_DECREF(tmp); + } + #endif + if (likely(v)) { + int one = 1; int is_little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&val; + int ret = _PyLong_AsByteArray((PyLongObject *)v, + bytes, sizeof(val), + is_little, !is_unsigned); + Py_DECREF(v); + if (likely(!ret)) + return val; + } +#endif + return (int) -1; + } + } else { + int val; + PyObject *tmp = __Pyx_PyNumber_IntOrLong(x); + if (!tmp) return (int) -1; + val = __Pyx_PyInt_As_int(tmp); + Py_DECREF(tmp); + return val; + } +raise_overflow: + PyErr_SetString(PyExc_OverflowError, + "value too large to convert to int"); + return (int) -1; +raise_neg_overflow: + PyErr_SetString(PyExc_OverflowError, + "can't convert negative value to int"); + return (int) -1; +} + +/* CIntToPy */ +static CYTHON_INLINE PyObject* __Pyx_PyInt_From_long(long value) { + const long neg_one = (long) ((long) 0 - (long) 1), const_zero = (long) 0; + const int is_unsigned = neg_one > const_zero; + if (is_unsigned) { + if (sizeof(long) < sizeof(long)) { + return PyInt_FromLong((long) value); + } else if (sizeof(long) <= sizeof(unsigned long)) { + return PyLong_FromUnsignedLong((unsigned long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(long) <= sizeof(unsigned PY_LONG_LONG)) { + return PyLong_FromUnsignedLongLong((unsigned PY_LONG_LONG) value); +#endif + } + } else { + if (sizeof(long) <= sizeof(long)) { + return PyInt_FromLong((long) value); +#ifdef HAVE_LONG_LONG + } else if (sizeof(long) <= sizeof(PY_LONG_LONG)) { + return PyLong_FromLongLong((PY_LONG_LONG) value); +#endif + } + } + { + int one = 1; int little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&value; + return _PyLong_FromByteArray(bytes, sizeof(long), + little, !is_unsigned); + } +} + +/* CIntFromPy */ +static CYTHON_INLINE long __Pyx_PyInt_As_long(PyObject *x) { + const long neg_one = (long) ((long) 0 - (long) 1), const_zero = (long) 0; + const int is_unsigned = neg_one > const_zero; +#if PY_MAJOR_VERSION < 3 + if (likely(PyInt_Check(x))) { + if (sizeof(long) < sizeof(long)) { + __PYX_VERIFY_RETURN_INT(long, long, PyInt_AS_LONG(x)) + } else { + long val = PyInt_AS_LONG(x); + if (is_unsigned && unlikely(val < 0)) { + goto raise_neg_overflow; + } + return (long) val; + } + } else +#endif + if (likely(PyLong_Check(x))) { + if (is_unsigned) { +#if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)x)->ob_digit; + switch (Py_SIZE(x)) { + case 0: return (long) 0; + case 1: __PYX_VERIFY_RETURN_INT(long, digit, digits[0]) + case 2: + if (8 * sizeof(long) > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, unsigned long, (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) >= 2 * PyLong_SHIFT) { + return (long) (((((long)digits[1]) << PyLong_SHIFT) | (long)digits[0])); + } + } + break; + case 3: + if (8 * sizeof(long) > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, unsigned long, (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) >= 3 * PyLong_SHIFT) { + return (long) (((((((long)digits[2]) << PyLong_SHIFT) | (long)digits[1]) << PyLong_SHIFT) | (long)digits[0])); + } + } + break; + case 4: + if (8 * sizeof(long) > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, unsigned long, (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) >= 4 * PyLong_SHIFT) { + return (long) (((((((((long)digits[3]) << PyLong_SHIFT) | (long)digits[2]) << PyLong_SHIFT) | (long)digits[1]) << PyLong_SHIFT) | (long)digits[0])); + } + } + break; + } +#endif +#if CYTHON_COMPILING_IN_CPYTHON + if (unlikely(Py_SIZE(x) < 0)) { + goto raise_neg_overflow; + } +#else + { + int result = PyObject_RichCompareBool(x, Py_False, Py_LT); + if (unlikely(result < 0)) + return (long) -1; + if (unlikely(result == 1)) + goto raise_neg_overflow; + } +#endif + if (sizeof(long) <= sizeof(unsigned long)) { + __PYX_VERIFY_RETURN_INT_EXC(long, unsigned long, PyLong_AsUnsignedLong(x)) +#ifdef HAVE_LONG_LONG + } else if (sizeof(long) <= sizeof(unsigned PY_LONG_LONG)) { + __PYX_VERIFY_RETURN_INT_EXC(long, unsigned PY_LONG_LONG, PyLong_AsUnsignedLongLong(x)) +#endif + } + } else { +#if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)x)->ob_digit; + switch (Py_SIZE(x)) { + case 0: return (long) 0; + case -1: __PYX_VERIFY_RETURN_INT(long, sdigit, (sdigit) (-(sdigit)digits[0])) + case 1: __PYX_VERIFY_RETURN_INT(long, digit, +digits[0]) + case -2: + if (8 * sizeof(long) - 1 > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, long, -(long) (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) - 1 > 2 * PyLong_SHIFT) { + return (long) (((long)-1)*(((((long)digits[1]) << PyLong_SHIFT) | (long)digits[0]))); + } + } + break; + case 2: + if (8 * sizeof(long) > 1 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 2 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, unsigned long, (((((unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) - 1 > 2 * PyLong_SHIFT) { + return (long) ((((((long)digits[1]) << PyLong_SHIFT) | (long)digits[0]))); + } + } + break; + case -3: + if (8 * sizeof(long) - 1 > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, long, -(long) (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) - 1 > 3 * PyLong_SHIFT) { + return (long) (((long)-1)*(((((((long)digits[2]) << PyLong_SHIFT) | (long)digits[1]) << PyLong_SHIFT) | (long)digits[0]))); + } + } + break; + case 3: + if (8 * sizeof(long) > 2 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 3 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, unsigned long, (((((((unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) - 1 > 3 * PyLong_SHIFT) { + return (long) ((((((((long)digits[2]) << PyLong_SHIFT) | (long)digits[1]) << PyLong_SHIFT) | (long)digits[0]))); + } + } + break; + case -4: + if (8 * sizeof(long) - 1 > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, long, -(long) (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) - 1 > 4 * PyLong_SHIFT) { + return (long) (((long)-1)*(((((((((long)digits[3]) << PyLong_SHIFT) | (long)digits[2]) << PyLong_SHIFT) | (long)digits[1]) << PyLong_SHIFT) | (long)digits[0]))); + } + } + break; + case 4: + if (8 * sizeof(long) > 3 * PyLong_SHIFT) { + if (8 * sizeof(unsigned long) > 4 * PyLong_SHIFT) { + __PYX_VERIFY_RETURN_INT(long, unsigned long, (((((((((unsigned long)digits[3]) << PyLong_SHIFT) | (unsigned long)digits[2]) << PyLong_SHIFT) | (unsigned long)digits[1]) << PyLong_SHIFT) | (unsigned long)digits[0]))) + } else if (8 * sizeof(long) - 1 > 4 * PyLong_SHIFT) { + return (long) ((((((((((long)digits[3]) << PyLong_SHIFT) | (long)digits[2]) << PyLong_SHIFT) | (long)digits[1]) << PyLong_SHIFT) | (long)digits[0]))); + } + } + break; + } +#endif + if (sizeof(long) <= sizeof(long)) { + __PYX_VERIFY_RETURN_INT_EXC(long, long, PyLong_AsLong(x)) +#ifdef HAVE_LONG_LONG + } else if (sizeof(long) <= sizeof(PY_LONG_LONG)) { + __PYX_VERIFY_RETURN_INT_EXC(long, PY_LONG_LONG, PyLong_AsLongLong(x)) +#endif + } + } + { +#if CYTHON_COMPILING_IN_PYPY && !defined(_PyLong_AsByteArray) + PyErr_SetString(PyExc_RuntimeError, + "_PyLong_AsByteArray() not available in PyPy, cannot convert large numbers"); +#else + long val; + PyObject *v = __Pyx_PyNumber_IntOrLong(x); + #if PY_MAJOR_VERSION < 3 + if (likely(v) && !PyLong_Check(v)) { + PyObject *tmp = v; + v = PyNumber_Long(tmp); + Py_DECREF(tmp); + } + #endif + if (likely(v)) { + int one = 1; int is_little = (int)*(unsigned char *)&one; + unsigned char *bytes = (unsigned char *)&val; + int ret = _PyLong_AsByteArray((PyLongObject *)v, + bytes, sizeof(val), + is_little, !is_unsigned); + Py_DECREF(v); + if (likely(!ret)) + return val; + } +#endif + return (long) -1; + } + } else { + long val; + PyObject *tmp = __Pyx_PyNumber_IntOrLong(x); + if (!tmp) return (long) -1; + val = __Pyx_PyInt_As_long(tmp); + Py_DECREF(tmp); + return val; + } +raise_overflow: + PyErr_SetString(PyExc_OverflowError, + "value too large to convert to long"); + return (long) -1; +raise_neg_overflow: + PyErr_SetString(PyExc_OverflowError, + "can't convert negative value to long"); + return (long) -1; +} + +/* FastTypeChecks */ +#if CYTHON_COMPILING_IN_CPYTHON +static int __Pyx_InBases(PyTypeObject *a, PyTypeObject *b) { + while (a) { + a = a->tp_base; + if (a == b) + return 1; + } + return b == &PyBaseObject_Type; +} +static CYTHON_INLINE int __Pyx_IsSubtype(PyTypeObject *a, PyTypeObject *b) { + PyObject *mro; + if (a == b) return 1; + mro = a->tp_mro; + if (likely(mro)) { + Py_ssize_t i, n; + n = PyTuple_GET_SIZE(mro); + for (i = 0; i < n; i++) { + if (PyTuple_GET_ITEM(mro, i) == (PyObject *)b) + return 1; + } + return 0; + } + return __Pyx_InBases(a, b); +} +#if PY_MAJOR_VERSION == 2 +static int __Pyx_inner_PyErr_GivenExceptionMatches2(PyObject *err, PyObject* exc_type1, PyObject* exc_type2) { + PyObject *exception, *value, *tb; + int res; + __Pyx_PyThreadState_declare + __Pyx_PyThreadState_assign + __Pyx_ErrFetch(&exception, &value, &tb); + res = exc_type1 ? PyObject_IsSubclass(err, exc_type1) : 0; + if (unlikely(res == -1)) { + PyErr_WriteUnraisable(err); + res = 0; + } + if (!res) { + res = PyObject_IsSubclass(err, exc_type2); + if (unlikely(res == -1)) { + PyErr_WriteUnraisable(err); + res = 0; + } + } + __Pyx_ErrRestore(exception, value, tb); + return res; +} +#else +static CYTHON_INLINE int __Pyx_inner_PyErr_GivenExceptionMatches2(PyObject *err, PyObject* exc_type1, PyObject *exc_type2) { + int res = exc_type1 ? __Pyx_IsSubtype((PyTypeObject*)err, (PyTypeObject*)exc_type1) : 0; + if (!res) { + res = __Pyx_IsSubtype((PyTypeObject*)err, (PyTypeObject*)exc_type2); + } + return res; +} +#endif +static int __Pyx_PyErr_GivenExceptionMatchesTuple(PyObject *exc_type, PyObject *tuple) { + Py_ssize_t i, n; + assert(PyExceptionClass_Check(exc_type)); + n = PyTuple_GET_SIZE(tuple); +#if PY_MAJOR_VERSION >= 3 + for (i=0; i= 0x02070000 + if (!PyCapsule_IsValid(cobj, sig)) { + PyErr_Format(PyExc_TypeError, + "C variable %.200s.%.200s has wrong signature (expected %.500s, got %.500s)", + PyModule_GetName(module), name, sig, PyCapsule_GetName(cobj)); + goto bad; + } + *p = PyCapsule_GetPointer(cobj, sig); +#else + {const char *desc, *s1, *s2; + desc = (const char *)PyCObject_GetDesc(cobj); + if (!desc) + goto bad; + s1 = desc; s2 = sig; + while (*s1 != '\0' && *s1 == *s2) { s1++; s2++; } + if (*s1 != *s2) { + PyErr_Format(PyExc_TypeError, + "C variable %.200s.%.200s has wrong signature (expected %.500s, got %.500s)", + PyModule_GetName(module), name, sig, desc); + goto bad; + } + *p = PyCObject_AsVoidPtr(cobj);} +#endif + if (!(*p)) + goto bad; + Py_DECREF(d); + return 0; +bad: + Py_XDECREF(d); + return -1; +} +#endif + +/* FunctionImport */ +#ifndef __PYX_HAVE_RT_ImportFunction +#define __PYX_HAVE_RT_ImportFunction +static int __Pyx_ImportFunction(PyObject *module, const char *funcname, void (**f)(void), const char *sig) { + PyObject *d = 0; + PyObject *cobj = 0; + union { + void (*fp)(void); + void *p; + } tmp; + d = PyObject_GetAttrString(module, (char *)"__pyx_capi__"); + if (!d) + goto bad; + cobj = PyDict_GetItemString(d, funcname); + if (!cobj) { + PyErr_Format(PyExc_ImportError, + "%.200s does not export expected C function %.200s", + PyModule_GetName(module), funcname); + goto bad; + } +#if PY_VERSION_HEX >= 0x02070000 + if (!PyCapsule_IsValid(cobj, sig)) { + PyErr_Format(PyExc_TypeError, + "C function %.200s.%.200s has wrong signature (expected %.500s, got %.500s)", + PyModule_GetName(module), funcname, sig, PyCapsule_GetName(cobj)); + goto bad; + } + tmp.p = PyCapsule_GetPointer(cobj, sig); +#else + {const char *desc, *s1, *s2; + desc = (const char *)PyCObject_GetDesc(cobj); + if (!desc) + goto bad; + s1 = desc; s2 = sig; + while (*s1 != '\0' && *s1 == *s2) { s1++; s2++; } + if (*s1 != *s2) { + PyErr_Format(PyExc_TypeError, + "C function %.200s.%.200s has wrong signature (expected %.500s, got %.500s)", + PyModule_GetName(module), funcname, sig, desc); + goto bad; + } + tmp.p = PyCObject_AsVoidPtr(cobj);} +#endif + *f = tmp.fp; + if (!(*f)) + goto bad; + Py_DECREF(d); + return 0; +bad: + Py_XDECREF(d); + return -1; +} +#endif + +/* InitStrings */ +static int __Pyx_InitStrings(__Pyx_StringTabEntry *t) { + while (t->p) { + #if PY_MAJOR_VERSION < 3 + if (t->is_unicode) { + *t->p = PyUnicode_DecodeUTF8(t->s, t->n - 1, NULL); + } else if (t->intern) { + *t->p = PyString_InternFromString(t->s); + } else { + *t->p = PyString_FromStringAndSize(t->s, t->n - 1); + } + #else + if (t->is_unicode | t->is_str) { + if (t->intern) { + *t->p = PyUnicode_InternFromString(t->s); + } else if (t->encoding) { + *t->p = PyUnicode_Decode(t->s, t->n - 1, t->encoding, NULL); + } else { + *t->p = PyUnicode_FromStringAndSize(t->s, t->n - 1); + } + } else { + *t->p = PyBytes_FromStringAndSize(t->s, t->n - 1); + } + #endif + if (!*t->p) + return -1; + if (PyObject_Hash(*t->p) == -1) + return -1; + ++t; + } + return 0; +} + +static CYTHON_INLINE PyObject* __Pyx_PyUnicode_FromString(const char* c_str) { + return __Pyx_PyUnicode_FromStringAndSize(c_str, (Py_ssize_t)strlen(c_str)); +} +static CYTHON_INLINE const char* __Pyx_PyObject_AsString(PyObject* o) { + Py_ssize_t ignore; + return __Pyx_PyObject_AsStringAndSize(o, &ignore); +} +#if __PYX_DEFAULT_STRING_ENCODING_IS_ASCII || __PYX_DEFAULT_STRING_ENCODING_IS_DEFAULT +#if !CYTHON_PEP393_ENABLED +static const char* __Pyx_PyUnicode_AsStringAndSize(PyObject* o, Py_ssize_t *length) { + char* defenc_c; + PyObject* defenc = _PyUnicode_AsDefaultEncodedString(o, NULL); + if (!defenc) return NULL; + defenc_c = PyBytes_AS_STRING(defenc); +#if __PYX_DEFAULT_STRING_ENCODING_IS_ASCII + { + char* end = defenc_c + PyBytes_GET_SIZE(defenc); + char* c; + for (c = defenc_c; c < end; c++) { + if ((unsigned char) (*c) >= 128) { + PyUnicode_AsASCIIString(o); + return NULL; + } + } + } +#endif + *length = PyBytes_GET_SIZE(defenc); + return defenc_c; +} +#else +static CYTHON_INLINE const char* __Pyx_PyUnicode_AsStringAndSize(PyObject* o, Py_ssize_t *length) { + if (unlikely(__Pyx_PyUnicode_READY(o) == -1)) return NULL; +#if __PYX_DEFAULT_STRING_ENCODING_IS_ASCII + if (likely(PyUnicode_IS_ASCII(o))) { + *length = PyUnicode_GET_LENGTH(o); + return PyUnicode_AsUTF8(o); + } else { + PyUnicode_AsASCIIString(o); + return NULL; + } +#else + return PyUnicode_AsUTF8AndSize(o, length); +#endif +} +#endif +#endif +static CYTHON_INLINE const char* __Pyx_PyObject_AsStringAndSize(PyObject* o, Py_ssize_t *length) { +#if __PYX_DEFAULT_STRING_ENCODING_IS_ASCII || __PYX_DEFAULT_STRING_ENCODING_IS_DEFAULT + if ( +#if PY_MAJOR_VERSION < 3 && __PYX_DEFAULT_STRING_ENCODING_IS_ASCII + __Pyx_sys_getdefaultencoding_not_ascii && +#endif + PyUnicode_Check(o)) { + return __Pyx_PyUnicode_AsStringAndSize(o, length); + } else +#endif +#if (!CYTHON_COMPILING_IN_PYPY) || (defined(PyByteArray_AS_STRING) && defined(PyByteArray_GET_SIZE)) + if (PyByteArray_Check(o)) { + *length = PyByteArray_GET_SIZE(o); + return PyByteArray_AS_STRING(o); + } else +#endif + { + char* result; + int r = PyBytes_AsStringAndSize(o, &result, length); + if (unlikely(r < 0)) { + return NULL; + } else { + return result; + } + } +} +static CYTHON_INLINE int __Pyx_PyObject_IsTrue(PyObject* x) { + int is_true = x == Py_True; + if (is_true | (x == Py_False) | (x == Py_None)) return is_true; + else return PyObject_IsTrue(x); +} +static CYTHON_INLINE int __Pyx_PyObject_IsTrueAndDecref(PyObject* x) { + int retval; + if (unlikely(!x)) return -1; + retval = __Pyx_PyObject_IsTrue(x); + Py_DECREF(x); + return retval; +} +static PyObject* __Pyx_PyNumber_IntOrLongWrongResultType(PyObject* result, const char* type_name) { +#if PY_MAJOR_VERSION >= 3 + if (PyLong_Check(result)) { + if (PyErr_WarnFormat(PyExc_DeprecationWarning, 1, + "__int__ returned non-int (type %.200s). " + "The ability to return an instance of a strict subclass of int " + "is deprecated, and may be removed in a future version of Python.", + Py_TYPE(result)->tp_name)) { + Py_DECREF(result); + return NULL; + } + return result; + } +#endif + PyErr_Format(PyExc_TypeError, + "__%.4s__ returned non-%.4s (type %.200s)", + type_name, type_name, Py_TYPE(result)->tp_name); + Py_DECREF(result); + return NULL; +} +static CYTHON_INLINE PyObject* __Pyx_PyNumber_IntOrLong(PyObject* x) { +#if CYTHON_USE_TYPE_SLOTS + PyNumberMethods *m; +#endif + const char *name = NULL; + PyObject *res = NULL; +#if PY_MAJOR_VERSION < 3 + if (likely(PyInt_Check(x) || PyLong_Check(x))) +#else + if (likely(PyLong_Check(x))) +#endif + return __Pyx_NewRef(x); +#if CYTHON_USE_TYPE_SLOTS + m = Py_TYPE(x)->tp_as_number; + #if PY_MAJOR_VERSION < 3 + if (m && m->nb_int) { + name = "int"; + res = m->nb_int(x); + } + else if (m && m->nb_long) { + name = "long"; + res = m->nb_long(x); + } + #else + if (likely(m && m->nb_int)) { + name = "int"; + res = m->nb_int(x); + } + #endif +#else + if (!PyBytes_CheckExact(x) && !PyUnicode_CheckExact(x)) { + res = PyNumber_Int(x); + } +#endif + if (likely(res)) { +#if PY_MAJOR_VERSION < 3 + if (unlikely(!PyInt_Check(res) && !PyLong_Check(res))) { +#else + if (unlikely(!PyLong_CheckExact(res))) { +#endif + return __Pyx_PyNumber_IntOrLongWrongResultType(res, name); + } + } + else if (!PyErr_Occurred()) { + PyErr_SetString(PyExc_TypeError, + "an integer is required"); + } + return res; +} +static CYTHON_INLINE Py_ssize_t __Pyx_PyIndex_AsSsize_t(PyObject* b) { + Py_ssize_t ival; + PyObject *x; +#if PY_MAJOR_VERSION < 3 + if (likely(PyInt_CheckExact(b))) { + if (sizeof(Py_ssize_t) >= sizeof(long)) + return PyInt_AS_LONG(b); + else + return PyInt_AsSsize_t(b); + } +#endif + if (likely(PyLong_CheckExact(b))) { + #if CYTHON_USE_PYLONG_INTERNALS + const digit* digits = ((PyLongObject*)b)->ob_digit; + const Py_ssize_t size = Py_SIZE(b); + if (likely(__Pyx_sst_abs(size) <= 1)) { + ival = likely(size) ? digits[0] : 0; + if (size == -1) ival = -ival; + return ival; + } else { + switch (size) { + case 2: + if (8 * sizeof(Py_ssize_t) > 2 * PyLong_SHIFT) { + return (Py_ssize_t) (((((size_t)digits[1]) << PyLong_SHIFT) | (size_t)digits[0])); + } + break; + case -2: + if (8 * sizeof(Py_ssize_t) > 2 * PyLong_SHIFT) { + return -(Py_ssize_t) (((((size_t)digits[1]) << PyLong_SHIFT) | (size_t)digits[0])); + } + break; + case 3: + if (8 * sizeof(Py_ssize_t) > 3 * PyLong_SHIFT) { + return (Py_ssize_t) (((((((size_t)digits[2]) << PyLong_SHIFT) | (size_t)digits[1]) << PyLong_SHIFT) | (size_t)digits[0])); + } + break; + case -3: + if (8 * sizeof(Py_ssize_t) > 3 * PyLong_SHIFT) { + return -(Py_ssize_t) (((((((size_t)digits[2]) << PyLong_SHIFT) | (size_t)digits[1]) << PyLong_SHIFT) | (size_t)digits[0])); + } + break; + case 4: + if (8 * sizeof(Py_ssize_t) > 4 * PyLong_SHIFT) { + return (Py_ssize_t) (((((((((size_t)digits[3]) << PyLong_SHIFT) | (size_t)digits[2]) << PyLong_SHIFT) | (size_t)digits[1]) << PyLong_SHIFT) | (size_t)digits[0])); + } + break; + case -4: + if (8 * sizeof(Py_ssize_t) > 4 * PyLong_SHIFT) { + return -(Py_ssize_t) (((((((((size_t)digits[3]) << PyLong_SHIFT) | (size_t)digits[2]) << PyLong_SHIFT) | (size_t)digits[1]) << PyLong_SHIFT) | (size_t)digits[0])); + } + break; + } + } + #endif + return PyLong_AsSsize_t(b); + } + x = PyNumber_Index(b); + if (!x) return -1; + ival = PyInt_AsSsize_t(x); + Py_DECREF(x); + return ival; +} +static CYTHON_INLINE PyObject * __Pyx_PyBool_FromLong(long b) { + return b ? __Pyx_NewRef(Py_True) : __Pyx_NewRef(Py_False); +} +static CYTHON_INLINE PyObject * __Pyx_PyInt_FromSize_t(size_t ival) { + return PyInt_FromSize_t(ival); +} + + +#endif /* Py_PYTHON_H */ diff --git a/fse/models/pooling_inner.pyx b/fse/models/pooling_inner.pyx new file mode 100644 index 0000000..d8000fd --- /dev/null +++ b/fse/models/pooling_inner.pyx @@ -0,0 +1,211 @@ +#!/usr/bin/env cython +# cython: boundscheck=False +# cython: wraparound=False +# cython: cdivision=True +# cython: embedsignature=True +# coding: utf-8 + +# Author: Oliver Borchers +# Copyright (C) 2019 Oliver Borchers + +"""Optimized cython functions for computing sentence embeddings""" + +import cython +import numpy as np + +cimport numpy as np + +from libc.string cimport memset + +import scipy.linalg.blas as fblas + +from average_inner cimport ( + REAL_t, + uINT_t, + ONE, + ZERO, + ONEF, + ZEROF, + saxpy, + sscal, + BaseSentenceVecsConfig, + FTSentenceVecsConfig, + init_base_s2v_config, + init_ft_s2v_config, + populate_base_s2v_config, + populate_ft_s2v_config, +) + +DEF MAX_WORDS = 10000 +DEF MAX_NGRAMS = 40 + +cdef void sl_max_pool(const int *N, float *X, const float *Y) nogil: + """ Performs single left max pooling op + + Parameters + ---------- + N : int * + Vector size. + X : float * + Left vector. + Y : float * + Right vector. + + """ + cdef int i + for i from 0 <= i < N[0] by 1: + if X[i] < Y[i]: + X[i] = Y[i] + +cdef void compute_base_sentence_pooling(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: + """Perform optimized sentence-level max pooling for BaseAny2Vec model. + + Parameters + ---------- + c : BaseSentenceVecsConfig * + A pointer to a fully initialized and populated struct. + num_sentences : uINT_t + The number of sentences used to train the model. + + Notes + ----- + This routine does not provide oov support. + + """ + cdef: + int size = c.size + + uINT_t sent_idx, sent_start, sent_end, sent_row + + uINT_t i, word_idx, word_row + + REAL_t sent_len, inv_count + + for sent_idx in range(num_sentences): + memset(c.mem, 0, size * cython.sizeof(REAL_t)) + memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + + sent_start = c.sentence_boundary[sent_idx] + sent_end = c.sentence_boundary[sent_idx + 1] + sent_len = ZEROF + + for i in range(sent_start, sent_end): + sent_len += ONEF + sent_row = c.sent_adresses[i] * size + word_row = c.word_indices[i] * size + + sl_max_pool( + &size, + &c.sentence_vectors[sent_row], + &c.word_vectors[word_row], + ) + # There's nothing to do here for many-to-one mappings + + +# cdef void compute_ft_sentence_averages(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: +# """Perform optimized sentence-level averaging for FastText model. + +# Parameters +# ---------- +# c : FTSentenceVecsConfig * +# A pointer to a fully initialized and populated struct. +# num_sentences : uINT_t +# The number of sentences used to train the model. + +# Notes +# ----- +# This routine DOES provide oov support. + +# """ +# cdef: +# int size = c.size + +# uINT_t sent_idx, sent_start, sent_end, sent_row + +# uINT_t ngram_row, ngrams + +# uINT_t i, j, word_idx, word_row + +# REAL_t sent_len +# REAL_t inv_count, inv_ngram +# REAL_t oov_weight = c.oov_weight + + +# for sent_idx in range(num_sentences): +# memset(c.mem, 0, size * cython.sizeof(REAL_t)) +# sent_start = c.sentence_boundary[sent_idx] +# sent_end = c.sentence_boundary[sent_idx + 1] +# sent_len = ZEROF + +# for i in range(sent_start, sent_end): +# sent_len += ONEF +# sent_row = c.sent_adresses[i] * size + +# word_idx = c.word_indices[i] +# ngrams = c.subwords_idx_len[i] + +# if ngrams == 0: +# word_row = c.word_indices[i] * size +# saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) +# else: +# inv_ngram = (ONEF / ngrams) * c.oov_weight +# for j in range(ngrams): +# ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size +# saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) + +# if sent_len > ZEROF: +# inv_count = ONEF / sent_len +# saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + +def train_pooling_cy(model, indexed_sentences, target, memory): + """Training on a sequence of sentences and update the target ndarray. + + Called internally from :meth:`~fse.models.pooling.MaxPooling._do_train_job`. + + Parameters + ---------- + model : :class:`~fse.models.base_s2v.BaseSentence2VecModel` + The BaseSentence2VecModel model instance. + indexed_sentences : iterable of tuple + The sentences used to train the model. + target : ndarray + The target ndarray. We use the index from indexed_sentences + to write into the corresponding row of target. + memory : ndarray + Private memory for each working thread. + + Returns + ------- + int, int + Number of effective sentences (non-zero) and effective words in the vocabulary used + during training the sentence embedding. + """ + + cdef uINT_t eff_sentences = 0 + cdef uINT_t eff_words = 0 + cdef BaseSentenceVecsConfig w2v + cdef FTSentenceVecsConfig ft + + if not model.is_ft: + init_base_s2v_config(&w2v, model, target, memory) + + eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + + with nogil: + compute_base_sentence_pooling(&w2v, eff_sentences) + else: + init_ft_s2v_config(&ft, model, target, memory) + + eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + + # with nogil: + # compute_ft_sentence_averages(&ft, eff_sentences) + + return eff_sentences, eff_words + +def init(): + return 1 + +MAX_WORDS_IN_BATCH = MAX_WORDS +MAX_NGRAMS_IN_BATCH = MAX_NGRAMS +FAST_VERSION = init() \ No newline at end of file diff --git a/fse/test/test_base_s2v.py b/fse/test/test_base_s2v.py index 4193b30..ed03687 100644 --- a/fse/test/test_base_s2v.py +++ b/fse/test/test_base_s2v.py @@ -428,6 +428,13 @@ def temp_train_job(data_iterable, target, memory): report_delay=0.01, ) self.assertEqual((100, 200, 300), job_output) + + def test_memory_creation(self): + se = BaseSentence2VecModel(W2V, workers=2) + out = se._get_thread_working_mem() + self.assertEqual(3, len(out)) + self.assertEqual(se.sv.vector_size, len(out[0])) + self.assertEqual(se.sv.vector_size, len(out[2])) def test_infer_method(self): se = BaseSentence2VecModel(W2V) diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index 519df98..37cd5bb 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -45,7 +45,7 @@ FT_R.build_vocab(SENTENCES) -class TestAverageFunctions(unittest.TestCase): +class TestPoolingFunctions(unittest.TestCase): def setUp(self): self.sentences = [ ["They", "admit"], @@ -61,32 +61,20 @@ def setUp(self): ) self.model._pre_train_calls() - # def test_cython(self): - # from fse.models.average_inner import ( - # FAST_VERSION, - # MAX_WORDS_IN_BATCH, - # MAX_NGRAMS_IN_BATCH, - # ) - - # self.assertTrue(FAST_VERSION) - # self.assertEqual(10000, MAX_WORDS_IN_BATCH) - # self.assertEqual(40, MAX_NGRAMS_IN_BATCH) - - # def test_average_train_cy_w2v(self): - # self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) - # mem = self.model._get_thread_working_mem() - - # from fse.models.average_inner import train_average_cy - - # output = train_average_cy( - # self.model, self.sentences, self.model.sv.vectors, mem - # ) - # self.assertEqual((4, 7), output) - # self.assertTrue((183 == self.model.sv[0]).all()) - # self.assertTrue((164.5 == self.model.sv[1]).all()) - # self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) + def test_cython(self): + from fse.models.pooling_inner import ( + FAST_VERSION, + MAX_WORDS_IN_BATCH, + MAX_NGRAMS_IN_BATCH, + train_pooling_cy, + ) + self.assertTrue(FAST_VERSION) + self.assertTrue(callable(train_pooling_cy)) + self.assertEqual(10000, MAX_WORDS_IN_BATCH) + self.assertEqual(40, MAX_NGRAMS_IN_BATCH) # def test_average_train_cy_ft(self): + # TODO # ft = FastText(min_count=1, size=DIM) # ft.build_vocab(SENTENCES) # m = Average(ft) @@ -106,56 +94,8 @@ def setUp(self): # self.assertTrue(np.allclose(1.5, m.sv[2])) # self.assertTrue(np.allclose(2, m.sv[3])) - # def test_cy_equal_np_w2v(self): - # m1 = Average(W2V) - # m1.prep.prepare_vectors( - # sv=m1.sv, total_sentences=len(self.sentences), update=False - # ) - # m1._pre_train_calls() - # mem1 = m1._get_thread_working_mem() - # o1 = train_average_np(m1, self.sentences, m1.sv.vectors, mem1) - - # m2 = Average(W2V) - # m2.prep.prepare_vectors( - # sv=m2.sv, total_sentences=len(self.sentences), update=False - # ) - # m2._pre_train_calls() - # mem2 = m2._get_thread_working_mem() - - # from fse.models.average_inner import train_average_cy - - # o2 = train_average_cy(m2, self.sentences, m2.sv.vectors, mem2) - - # self.assertEqual(o1, o2) - # self.assertTrue((m1.sv.vectors == m2.sv.vectors).all()) - - # def test_cy_equal_np_w2v_random(self): - # w2v = Word2Vec(min_count=1, size=DIM) - # # Random initialization - # w2v.build_vocab(SENTENCES) - - # m1 = Average(w2v) - # m1.prep.prepare_vectors( - # sv=m1.sv, total_sentences=len(self.sentences), update=False - # ) - # m1._pre_train_calls() - # mem1 = m1._get_thread_working_mem() - # o1 = train_average_np(m1, self.sentences, m1.sv.vectors, mem1) - - # m2 = Average(w2v) - # m2.prep.prepare_vectors( - # sv=m2.sv, total_sentences=len(self.sentences), update=False - # ) - # m2._pre_train_calls() - # mem2 = m2._get_thread_working_mem() - - # from fse.models.average_inner import train_average_cy - - # o2 = train_average_cy(m2, self.sentences, m2.sv.vectors, mem2) - - # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - # def test_cy_equal_np_ft_random(self): + # TODO # ft = FastText(size=20, min_count=1) # ft.build_vocab(SENTENCES) @@ -186,6 +126,7 @@ def setUp(self): # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) # def test_train_single_from_disk(self): + # TODO # p = Path("fse/test/test_data/test_vecs") # p_res = Path("fse/test/test_data/test_vecs.vectors") # p_target = Path("fse/test/test_data/test_vecs_wv.vectors") @@ -206,6 +147,7 @@ def setUp(self): # p_target.unlink() # def test_train_multi_from_disk(self): + # TODO # p = Path("fse/test/test_data/test_vecs") # p_res = Path("fse/test/test_data/test_vecs.vectors") # p_target = Path("fse/test/test_data/test_vecs_wv.vectors") @@ -271,6 +213,70 @@ def test_pooling_train_np_w2v(self): self.assertTrue((306 == self.model.sv[1]).all()) self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) + def test_pool_train_cy_w2v(self): + self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + mem = self.model._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + output = train_pooling_cy( + self.model, self.sentences, self.model.sv.vectors, mem + ) + + self.assertEqual((5, 14), output) + self.assertTrue((241 == self.model.sv[0]).all()) + self.assertTrue((306 == self.model.sv[1]).all()) + self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) + + def test_cy_equal_np_w2v(self): + m1 = MaxPooling(W2V) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + + m2 = MaxPooling(W2V) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + + self.assertEqual(o1, o2) + self.assertTrue((m1.sv.vectors == m2.sv.vectors).all()) + + def test_cy_equal_np_w2v_random(self): + w2v = Word2Vec(min_count=1, size=DIM) + # Random initialization + w2v.build_vocab(SENTENCES) + + m1 = MaxPooling(w2v) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + + m2 = MaxPooling(w2v) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + + self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + def test_pooling_train_np_w2v_non_negative(self): mpool = MaxPooling(W2V_R) mpool.train(self.sentences) From 5feb9f3145b25a12628c1d1cc9234b10275aa3e8 Mon Sep 17 00:00:00 2001 From: OliverB Date: Thu, 27 Feb 2020 10:03:34 +0100 Subject: [PATCH 21/63] Working MaxPooling w2v & ft --- fse/models/base_s2v.py | 13 +- fse/models/pooling_inner.c | 567 ++++++++++++++++++++++++++++------- fse/models/pooling_inner.pyx | 114 ++++--- fse/test/test_pooling.py | 186 +++++------- 4 files changed, 603 insertions(+), 277 deletions(-) diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index 673255d..5fde303 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -132,7 +132,7 @@ def __init__( [ ] documentation [ ] more benchmarks [ ] more speed benchmarks -> worker support still somewhat weird - [ ] remove wv_mapfile_path? + [-] remove wv_mapfile_path? [ ] modifiable sv_mapfile_path? [ ] models: @@ -147,11 +147,12 @@ def __init__( [ ] sentencevectors: [X] similar_by_sentence model type check [ ] approximate NN search for large files - [ ] compare ANN libraries - [ ] ease-of-use - [ ] dependencies - [ ] compatibility - [ ] memory-usage + [X] compare ANN libraries + [X] ease-of-use + [X] dependencies + [X] compatibility + [X] memory-usage + --> Annoy """ set_madvise_for_mmap() diff --git a/fse/models/pooling_inner.c b/fse/models/pooling_inner.c index d8a776c..d8a72d0 100644 --- a/fse/models/pooling_inner.c +++ b/fse/models/pooling_inner.c @@ -1725,6 +1725,7 @@ static PyObject *(*__pyx_f_13average_inner_populate_ft_s2v_config)(struct __pyx_ /* Module declarations from 'pooling_inner' */ static void __pyx_f_13pooling_inner_sl_max_pool(int const *, float *, float const *); /*proto*/ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ +static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ #define __Pyx_MODULE_NAME "pooling_inner" extern int __pyx_module_is_main_pooling_inner; int __pyx_module_is_main_pooling_inner = 0; @@ -1757,6 +1758,7 @@ static const char __pyx_k_ValueError[] = "ValueError"; static const char __pyx_k_ImportError[] = "ImportError"; static const char __pyx_k_FAST_VERSION[] = "FAST_VERSION"; static const char __pyx_k_RuntimeError[] = "RuntimeError"; +static const char __pyx_k_hierarchical[] = "hierarchical"; static const char __pyx_k_eff_sentences[] = "eff_sentences"; static const char __pyx_k_pooling_inner[] = "pooling_inner"; static const char __pyx_k_train_pooling_cy[] = "train_pooling_cy"; @@ -1790,6 +1792,7 @@ static PyObject *__pyx_n_s_eff_sentences; static PyObject *__pyx_n_s_eff_words; static PyObject *__pyx_n_s_fblas; static PyObject *__pyx_n_s_ft; +static PyObject *__pyx_n_s_hierarchical; static PyObject *__pyx_n_s_import; static PyObject *__pyx_n_s_indexed_sentences; static PyObject *__pyx_n_s_init; @@ -2050,8 +2053,263 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t /* function exit code */ } -/* "pooling_inner.pyx":160 - * # saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) +/* "pooling_inner.pyx":105 + * + * + * cdef void compute_ft_sentence_pooling(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< + * """Perform optimized sentence-level max pooling for FastText model. + * + */ + +static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { + int __pyx_v_size; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_end; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; + __pyx_t_13average_inner_uINT_t __pyx_v_ngram_row; + __pyx_t_13average_inner_uINT_t __pyx_v_ngrams; + __pyx_t_13average_inner_uINT_t __pyx_v_i; + __pyx_t_13average_inner_uINT_t __pyx_v_j; + CYTHON_UNUSED __pyx_t_13average_inner_uINT_t __pyx_v_word_idx; + __pyx_t_13average_inner_uINT_t __pyx_v_word_row; + __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; + __pyx_t_13average_inner_REAL_t __pyx_v_inv_ngram; + CYTHON_UNUSED __pyx_t_13average_inner_REAL_t __pyx_v_oov_weight; + int __pyx_t_1; + __pyx_t_13average_inner_REAL_t __pyx_t_2; + __pyx_t_13average_inner_uINT_t __pyx_t_3; + __pyx_t_13average_inner_uINT_t __pyx_t_4; + __pyx_t_13average_inner_uINT_t __pyx_t_5; + __pyx_t_13average_inner_uINT_t __pyx_t_6; + __pyx_t_13average_inner_uINT_t __pyx_t_7; + __pyx_t_13average_inner_uINT_t __pyx_t_8; + int __pyx_t_9; + __pyx_t_13average_inner_uINT_t __pyx_t_10; + __pyx_t_13average_inner_uINT_t __pyx_t_11; + __pyx_t_13average_inner_uINT_t __pyx_t_12; + + /* "pooling_inner.pyx":121 + * """ + * cdef: + * int size = c.size # <<<<<<<<<<<<<< + * + * uINT_t sent_idx, sent_start, sent_end, sent_row + */ + __pyx_t_1 = __pyx_v_c->size; + __pyx_v_size = __pyx_t_1; + + /* "pooling_inner.pyx":131 + * REAL_t sent_len + * REAL_t inv_count, inv_ngram + * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< + * + * for sent_idx in range(num_sentences): + */ + __pyx_t_2 = __pyx_v_c->oov_weight; + __pyx_v_oov_weight = __pyx_t_2; + + /* "pooling_inner.pyx":133 + * REAL_t oov_weight = c.oov_weight + * + * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] + */ + __pyx_t_3 = __pyx_v_num_sentences; + __pyx_t_4 = __pyx_t_3; + for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { + __pyx_v_sent_idx = __pyx_t_5; + + /* "pooling_inner.pyx":134 + * + * for sent_idx in range(num_sentences): + * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< + * sent_end = c.sentence_boundary[sent_idx + 1] + * sent_len = ZEROF + */ + __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); + + /* "pooling_inner.pyx":135 + * for sent_idx in range(num_sentences): + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< + * sent_len = ZEROF + * + */ + __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); + + /* "pooling_inner.pyx":136 + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] + * sent_len = ZEROF # <<<<<<<<<<<<<< + * + * for i in range(sent_start, sent_end): + */ + __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; + + /* "pooling_inner.pyx":138 + * sent_len = ZEROF + * + * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< + * sent_len += ONEF + * sent_row = c.sent_adresses[i] * size + */ + __pyx_t_6 = __pyx_v_sent_end; + __pyx_t_7 = __pyx_t_6; + for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { + __pyx_v_i = __pyx_t_8; + + /* "pooling_inner.pyx":139 + * + * for i in range(sent_start, sent_end): + * sent_len += ONEF # <<<<<<<<<<<<<< + * sent_row = c.sent_adresses[i] * size + * + */ + __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); + + /* "pooling_inner.pyx":140 + * for i in range(sent_start, sent_end): + * sent_len += ONEF + * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< + * + * word_idx = c.word_indices[i] + */ + __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); + + /* "pooling_inner.pyx":142 + * sent_row = c.sent_adresses[i] * size + * + * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< + * ngrams = c.subwords_idx_len[i] + * + */ + __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); + + /* "pooling_inner.pyx":143 + * + * word_idx = c.word_indices[i] + * ngrams = c.subwords_idx_len[i] # <<<<<<<<<<<<<< + * + * if ngrams == 0: + */ + __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_i]); + + /* "pooling_inner.pyx":145 + * ngrams = c.subwords_idx_len[i] + * + * if ngrams == 0: # <<<<<<<<<<<<<< + * word_row = c.word_indices[i] * size + * + */ + __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); + if (__pyx_t_9) { + + /* "pooling_inner.pyx":146 + * + * if ngrams == 0: + * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< + * + * sl_max_pool( + */ + __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); + + /* "pooling_inner.pyx":148 + * word_row = c.word_indices[i] * size + * + * sl_max_pool( # <<<<<<<<<<<<<< + * &size, + * &c.sentence_vectors[sent_row], + */ + __pyx_f_13pooling_inner_sl_max_pool((&__pyx_v_size), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row]))); + + /* "pooling_inner.pyx":145 + * ngrams = c.subwords_idx_len[i] + * + * if ngrams == 0: # <<<<<<<<<<<<<< + * word_row = c.word_indices[i] * size + * + */ + goto __pyx_L7; + } + + /* "pooling_inner.pyx":155 + * + * else: + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< + * inv_ngram = (ONEF / ngrams) * c.oov_weight + * for j in range(ngrams): + */ + /*else*/ { + (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); + + /* "pooling_inner.pyx":156 + * else: + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< + * for j in range(ngrams): + * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size + */ + __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); + + /* "pooling_inner.pyx":157 + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * inv_ngram = (ONEF / ngrams) * c.oov_weight + * for j in range(ngrams): # <<<<<<<<<<<<<< + * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size + * saxpy( + */ + __pyx_t_10 = __pyx_v_ngrams; + __pyx_t_11 = __pyx_t_10; + for (__pyx_t_12 = 0; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { + __pyx_v_j = __pyx_t_12; + + /* "pooling_inner.pyx":158 + * inv_ngram = (ONEF / ngrams) * c.oov_weight + * for j in range(ngrams): + * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size # <<<<<<<<<<<<<< + * saxpy( + * &size, + */ + __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_i * 40) + __pyx_v_j)]) * __pyx_v_size); + + /* "pooling_inner.pyx":159 + * for j in range(ngrams): + * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size + * saxpy( # <<<<<<<<<<<<<< + * &size, + * &inv_ngram, + */ + __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_ngram), (&(__pyx_v_c->ngram_vectors[__pyx_v_ngram_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); + } + + /* "pooling_inner.pyx":168 + * ) + * + * sl_max_pool( # <<<<<<<<<<<<<< + * &size, + * &c.sentence_vectors[sent_row], + */ + __pyx_f_13pooling_inner_sl_max_pool((&__pyx_v_size), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), __pyx_v_c->mem); + } + __pyx_L7:; + } + } + + /* "pooling_inner.pyx":105 + * + * + * cdef void compute_ft_sentence_pooling(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< + * """Perform optimized sentence-level max pooling for FastText model. + * + */ + + /* function exit code */ +} + +/* "pooling_inner.pyx":176 + * * * def train_pooling_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< * """Training on a sequence of sentences and update the target ndarray. @@ -2097,23 +2355,23 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self case 1: if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 160, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 176, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 2: if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 160, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 176, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 3: if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 160, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 176, __pyx_L3_error) } } if (unlikely(kw_args > 0)) { - if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 160, __pyx_L3_error) + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 176, __pyx_L3_error) } } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { goto __pyx_L5_argtuple_error; @@ -2130,7 +2388,7 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self } goto __pyx_L4_argument_unpacking_done; __pyx_L5_argtuple_error:; - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 160, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 176, __pyx_L3_error) __pyx_L3_error:; __Pyx_AddTraceback("pooling_inner.train_pooling_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __Pyx_RefNannyFinishContext(); @@ -2161,7 +2419,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_13average_inner_uINT_t __pyx_t_9; __Pyx_RefNannySetupContext("train_pooling_cy", 0); - /* "pooling_inner.pyx":184 + /* "pooling_inner.pyx":200 * """ * * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< @@ -2170,7 +2428,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_sentences = 0; - /* "pooling_inner.pyx":185 + /* "pooling_inner.pyx":201 * * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< @@ -2179,44 +2437,44 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_words = 0; - /* "pooling_inner.pyx":189 + /* "pooling_inner.pyx":205 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< * init_base_s2v_config(&w2v, model, target, memory) * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 189, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 205, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 189, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 205, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_3 = ((!__pyx_t_2) != 0); if (__pyx_t_3) { - /* "pooling_inner.pyx":190 + /* "pooling_inner.pyx":206 * * if not model.is_ft: * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< * * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) */ - __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 190, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 206, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":192 + /* "pooling_inner.pyx":208 * init_base_s2v_config(&w2v, model, target, memory) * * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) # <<<<<<<<<<<<<< * - * with nogil: + * if not model.hierarchical: */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 192, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 208, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 192, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 208, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 192, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 208, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -2225,7 +2483,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 192, __pyx_L1_error) + __PYX_ERR(0, 208, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -2238,15 +2496,15 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_4); __Pyx_INCREF(__pyx_t_5); #else - __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 192, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 208, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 192, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 208, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 192, __pyx_L1_error) + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 208, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; @@ -2254,7 +2512,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_4); index = 1; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L4_unpacking_failed; __Pyx_GOTREF(__pyx_t_5); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 192, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 208, __pyx_L1_error) __pyx_t_7 = NULL; __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; goto __pyx_L5_unpacking_done; @@ -2262,61 +2520,84 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_t_7 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 192, __pyx_L1_error) + __PYX_ERR(0, 208, __pyx_L1_error) __pyx_L5_unpacking_done:; } - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 192, __pyx_L1_error) + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 208, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 192, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 208, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; __pyx_v_eff_sentences = __pyx_t_8; __pyx_v_eff_words = __pyx_t_9; - /* "pooling_inner.pyx":194 + /* "pooling_inner.pyx":210 * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) * - * with nogil: # <<<<<<<<<<<<<< - * compute_base_sentence_pooling(&w2v, eff_sentences) - * else: + * if not model.hierarchical: # <<<<<<<<<<<<<< + * with nogil: + * compute_base_sentence_pooling(&w2v, eff_sentences) */ - { - #ifdef WITH_THREAD - PyThreadState *_save; - Py_UNBLOCK_THREADS - __Pyx_FastGIL_Remember(); - #endif - /*try:*/ { + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 210, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 210, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_2 = ((!__pyx_t_3) != 0); + if (__pyx_t_2) { - /* "pooling_inner.pyx":195 + /* "pooling_inner.pyx":211 * - * with nogil: - * compute_base_sentence_pooling(&w2v, eff_sentences) # <<<<<<<<<<<<<< + * if not model.hierarchical: + * with nogil: # <<<<<<<<<<<<<< + * compute_base_sentence_pooling(&w2v, eff_sentences) + * else: + */ + { + #ifdef WITH_THREAD + PyThreadState *_save; + Py_UNBLOCK_THREADS + __Pyx_FastGIL_Remember(); + #endif + /*try:*/ { + + /* "pooling_inner.pyx":212 + * if not model.hierarchical: + * with nogil: + * compute_base_sentence_pooling(&w2v, eff_sentences) # <<<<<<<<<<<<<< * else: * init_ft_s2v_config(&ft, model, target, memory) */ - __pyx_f_13pooling_inner_compute_base_sentence_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences); - } + __pyx_f_13pooling_inner_compute_base_sentence_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences); + } - /* "pooling_inner.pyx":194 - * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + /* "pooling_inner.pyx":211 * - * with nogil: # <<<<<<<<<<<<<< - * compute_base_sentence_pooling(&w2v, eff_sentences) + * if not model.hierarchical: + * with nogil: # <<<<<<<<<<<<<< + * compute_base_sentence_pooling(&w2v, eff_sentences) * else: */ - /*finally:*/ { - /*normal exit:*/{ - #ifdef WITH_THREAD - __Pyx_FastGIL_Forget(); - Py_BLOCK_THREADS - #endif - goto __pyx_L8; + /*finally:*/ { + /*normal exit:*/{ + #ifdef WITH_THREAD + __Pyx_FastGIL_Forget(); + Py_BLOCK_THREADS + #endif + goto __pyx_L9; + } + __pyx_L9:; } - __pyx_L8:; - } + } + + /* "pooling_inner.pyx":210 + * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + * + * if not model.hierarchical: # <<<<<<<<<<<<<< + * with nogil: + * compute_base_sentence_pooling(&w2v, eff_sentences) + */ } - /* "pooling_inner.pyx":189 + /* "pooling_inner.pyx":205 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< @@ -2326,31 +2607,31 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L3; } - /* "pooling_inner.pyx":197 - * compute_base_sentence_pooling(&w2v, eff_sentences) + /* "pooling_inner.pyx":214 + * compute_base_sentence_pooling(&w2v, eff_sentences) * else: * init_ft_s2v_config(&ft, model, target, memory) # <<<<<<<<<<<<<< * * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) */ /*else*/ { - __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 197, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 214, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":199 + /* "pooling_inner.pyx":216 * init_ft_s2v_config(&ft, model, target, memory) * * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) # <<<<<<<<<<<<<< * - * # with nogil: + * if not model.hierarchical: */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 199, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 216, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 199, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 216, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 199, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 216, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -2359,7 +2640,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 199, __pyx_L1_error) + __PYX_ERR(0, 216, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -2372,55 +2653,122 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_5); __Pyx_INCREF(__pyx_t_4); #else - __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 199, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 216, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); - __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 199, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 216, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 199, __pyx_L1_error) + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 216, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; - index = 0; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L9_unpacking_failed; + index = 0; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L10_unpacking_failed; __Pyx_GOTREF(__pyx_t_5); - index = 1; __pyx_t_4 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_4)) goto __pyx_L9_unpacking_failed; + index = 1; __pyx_t_4 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_4)) goto __pyx_L10_unpacking_failed; __Pyx_GOTREF(__pyx_t_4); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 199, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 216, __pyx_L1_error) __pyx_t_7 = NULL; __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; - goto __pyx_L10_unpacking_done; - __pyx_L9_unpacking_failed:; + goto __pyx_L11_unpacking_done; + __pyx_L10_unpacking_failed:; __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_t_7 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 199, __pyx_L1_error) - __pyx_L10_unpacking_done:; + __PYX_ERR(0, 216, __pyx_L1_error) + __pyx_L11_unpacking_done:; } - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 199, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 216, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 199, __pyx_L1_error) + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 216, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; __pyx_v_eff_sentences = __pyx_t_9; __pyx_v_eff_words = __pyx_t_8; + + /* "pooling_inner.pyx":218 + * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + * + * if not model.hierarchical: # <<<<<<<<<<<<<< + * with nogil: + * compute_ft_sentence_pooling(&ft, eff_sentences) + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 218, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 218, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_t_3 = ((!__pyx_t_2) != 0); + if (__pyx_t_3) { + + /* "pooling_inner.pyx":219 + * + * if not model.hierarchical: + * with nogil: # <<<<<<<<<<<<<< + * compute_ft_sentence_pooling(&ft, eff_sentences) + * + */ + { + #ifdef WITH_THREAD + PyThreadState *_save; + Py_UNBLOCK_THREADS + __Pyx_FastGIL_Remember(); + #endif + /*try:*/ { + + /* "pooling_inner.pyx":220 + * if not model.hierarchical: + * with nogil: + * compute_ft_sentence_pooling(&ft, eff_sentences) # <<<<<<<<<<<<<< + * + * return eff_sentences, eff_words + */ + __pyx_f_13pooling_inner_compute_ft_sentence_pooling((&__pyx_v_ft), __pyx_v_eff_sentences); + } + + /* "pooling_inner.pyx":219 + * + * if not model.hierarchical: + * with nogil: # <<<<<<<<<<<<<< + * compute_ft_sentence_pooling(&ft, eff_sentences) + * + */ + /*finally:*/ { + /*normal exit:*/{ + #ifdef WITH_THREAD + __Pyx_FastGIL_Forget(); + Py_BLOCK_THREADS + #endif + goto __pyx_L15; + } + __pyx_L15:; + } + } + + /* "pooling_inner.pyx":218 + * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + * + * if not model.hierarchical: # <<<<<<<<<<<<<< + * with nogil: + * compute_ft_sentence_pooling(&ft, eff_sentences) + */ + } } __pyx_L3:; - /* "pooling_inner.pyx":204 - * # compute_ft_sentence_averages(&ft, eff_sentences) + /* "pooling_inner.pyx":222 + * compute_ft_sentence_pooling(&ft, eff_sentences) * * return eff_sentences, eff_words # <<<<<<<<<<<<<< * * def init(): */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 204, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 222, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 204, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 222, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 204, __pyx_L1_error) + __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 222, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_1); @@ -2432,8 +2780,8 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_5 = 0; goto __pyx_L0; - /* "pooling_inner.pyx":160 - * # saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + /* "pooling_inner.pyx":176 + * * * def train_pooling_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< * """Training on a sequence of sentences and update the target ndarray. @@ -2454,7 +2802,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec return __pyx_r; } -/* "pooling_inner.pyx":206 +/* "pooling_inner.pyx":224 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -2482,7 +2830,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("init", 0); - /* "pooling_inner.pyx":207 + /* "pooling_inner.pyx":225 * * def init(): * return 1 # <<<<<<<<<<<<<< @@ -2494,7 +2842,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __pyx_r = __pyx_int_1; goto __pyx_L0; - /* "pooling_inner.pyx":206 + /* "pooling_inner.pyx":224 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -4988,6 +5336,7 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = { {&__pyx_n_s_eff_words, __pyx_k_eff_words, sizeof(__pyx_k_eff_words), 0, 0, 1, 1}, {&__pyx_n_s_fblas, __pyx_k_fblas, sizeof(__pyx_k_fblas), 0, 0, 1, 1}, {&__pyx_n_s_ft, __pyx_k_ft, sizeof(__pyx_k_ft), 0, 0, 1, 1}, + {&__pyx_n_s_hierarchical, __pyx_k_hierarchical, sizeof(__pyx_k_hierarchical), 0, 0, 1, 1}, {&__pyx_n_s_import, __pyx_k_import, sizeof(__pyx_k_import), 0, 0, 1, 1}, {&__pyx_n_s_indexed_sentences, __pyx_k_indexed_sentences, sizeof(__pyx_k_indexed_sentences), 0, 0, 1, 1}, {&__pyx_n_s_init, __pyx_k_init, sizeof(__pyx_k_init), 0, 0, 1, 1}, @@ -5106,26 +5455,26 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { __Pyx_GOTREF(__pyx_tuple__7); __Pyx_GIVEREF(__pyx_tuple__7); - /* "pooling_inner.pyx":160 - * # saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + /* "pooling_inner.pyx":176 + * * * def train_pooling_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< * """Training on a sequence of sentences and update the target ndarray. * */ - __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 160, __pyx_L1_error) + __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 176, __pyx_L1_error) __Pyx_GOTREF(__pyx_tuple__9); __Pyx_GIVEREF(__pyx_tuple__9); - __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 160, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 160, __pyx_L1_error) + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 176, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 176, __pyx_L1_error) - /* "pooling_inner.pyx":206 + /* "pooling_inner.pyx":224 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 206, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 206, __pyx_L1_error) + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 224, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 224, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -5488,58 +5837,58 @@ if (!__Pyx_RefNanny) { if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_2) < 0) __PYX_ERR(0, 20, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":160 - * # saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + /* "pooling_inner.pyx":176 + * * * def train_pooling_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< * """Training on a sequence of sentences and update the target ndarray. * */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 160, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 176, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 160, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 176, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":206 + /* "pooling_inner.pyx":224 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 206, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 224, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 206, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 224, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":209 + /* "pooling_inner.pyx":227 * return 1 * * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 209, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 227, __pyx_L1_error) - /* "pooling_inner.pyx":210 + /* "pooling_inner.pyx":228 * * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 210, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 228, __pyx_L1_error) - /* "pooling_inner.pyx":211 + /* "pooling_inner.pyx":229 * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() # <<<<<<<<<<<<<< */ - __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 211, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 229, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 211, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 229, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 211, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 229, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; /* "pooling_inner.pyx":1 diff --git a/fse/models/pooling_inner.pyx b/fse/models/pooling_inner.pyx index d8000fd..2c7313b 100644 --- a/fse/models/pooling_inner.pyx +++ b/fse/models/pooling_inner.pyx @@ -102,60 +102,76 @@ cdef void compute_base_sentence_pooling(BaseSentenceVecsConfig *c, uINT_t num_se # There's nothing to do here for many-to-one mappings -# cdef void compute_ft_sentence_averages(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: -# """Perform optimized sentence-level averaging for FastText model. - -# Parameters -# ---------- -# c : FTSentenceVecsConfig * -# A pointer to a fully initialized and populated struct. -# num_sentences : uINT_t -# The number of sentences used to train the model. - -# Notes -# ----- -# This routine DOES provide oov support. +cdef void compute_ft_sentence_pooling(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: + """Perform optimized sentence-level max pooling for FastText model. -# """ -# cdef: -# int size = c.size + Parameters + ---------- + c : FTSentenceVecsConfig * + A pointer to a fully initialized and populated struct. + num_sentences : uINT_t + The number of sentences used to train the model. + + Notes + ----- + This routine DOES provide oov support. -# uINT_t sent_idx, sent_start, sent_end, sent_row + """ + cdef: + int size = c.size -# uINT_t ngram_row, ngrams + uINT_t sent_idx, sent_start, sent_end, sent_row -# uINT_t i, j, word_idx, word_row + uINT_t ngram_row, ngrams -# REAL_t sent_len -# REAL_t inv_count, inv_ngram -# REAL_t oov_weight = c.oov_weight + uINT_t i, j, word_idx, word_row + REAL_t sent_len + REAL_t inv_count, inv_ngram + REAL_t oov_weight = c.oov_weight -# for sent_idx in range(num_sentences): -# memset(c.mem, 0, size * cython.sizeof(REAL_t)) -# sent_start = c.sentence_boundary[sent_idx] -# sent_end = c.sentence_boundary[sent_idx + 1] -# sent_len = ZEROF + for sent_idx in range(num_sentences): + sent_start = c.sentence_boundary[sent_idx] + sent_end = c.sentence_boundary[sent_idx + 1] + sent_len = ZEROF -# for i in range(sent_start, sent_end): -# sent_len += ONEF -# sent_row = c.sent_adresses[i] * size + for i in range(sent_start, sent_end): + sent_len += ONEF + sent_row = c.sent_adresses[i] * size -# word_idx = c.word_indices[i] -# ngrams = c.subwords_idx_len[i] + word_idx = c.word_indices[i] + ngrams = c.subwords_idx_len[i] + + if ngrams == 0: + word_row = c.word_indices[i] * size + + sl_max_pool( + &size, + &c.sentence_vectors[sent_row], + &c.word_vectors[word_row], + ) + + else: + memset(c.mem, 0, size * cython.sizeof(REAL_t)) + inv_ngram = (ONEF / ngrams) * c.oov_weight + for j in range(ngrams): + ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size + saxpy( + &size, + &inv_ngram, + &c.ngram_vectors[ngram_row], + &ONE, + c.mem, + &ONE + ) + + sl_max_pool( + &size, + &c.sentence_vectors[sent_row], + c.mem, + ) + # There's nothing to do here for many-to-one mappings -# if ngrams == 0: -# word_row = c.word_indices[i] * size -# saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) -# else: -# inv_ngram = (ONEF / ngrams) * c.oov_weight -# for j in range(ngrams): -# ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size -# saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) - -# if sent_len > ZEROF: -# inv_count = ONEF / sent_len -# saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) def train_pooling_cy(model, indexed_sentences, target, memory): """Training on a sequence of sentences and update the target ndarray. @@ -191,15 +207,17 @@ def train_pooling_cy(model, indexed_sentences, target, memory): eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) - with nogil: - compute_base_sentence_pooling(&w2v, eff_sentences) + if not model.hierarchical: + with nogil: + compute_base_sentence_pooling(&w2v, eff_sentences) else: init_ft_s2v_config(&ft, model, target, memory) eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) - # with nogil: - # compute_ft_sentence_averages(&ft, eff_sentences) + if not model.hierarchical: + with nogil: + compute_ft_sentence_pooling(&ft, eff_sentences) return eff_sentences, eff_words diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index 37cd5bb..e477851 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -73,103 +73,6 @@ def test_cython(self): self.assertEqual(10000, MAX_WORDS_IN_BATCH) self.assertEqual(40, MAX_NGRAMS_IN_BATCH) - # def test_average_train_cy_ft(self): - # TODO - # ft = FastText(min_count=1, size=DIM) - # ft.build_vocab(SENTENCES) - # m = Average(ft) - # m.prep.prepare_vectors( - # sv=m.sv, total_sentences=len(self.sentences), update=False - # ) - # m._pre_train_calls() - # m.wv.vectors = m.wv.vectors_vocab = np.ones_like(m.wv.vectors, dtype=np.float32) - # m.wv.vectors_ngrams = np.full_like(m.wv.vectors_ngrams, 2, dtype=np.float32) - # mem = m._get_thread_working_mem() - - # from fse.models.average_inner import train_average_cy - - # output = train_average_cy(m, self.sentences, m.sv.vectors, mem) - # self.assertEqual((4, 10), output) - # self.assertTrue((1.0 + EPS == m.sv[0]).all()) - # self.assertTrue(np.allclose(1.5, m.sv[2])) - # self.assertTrue(np.allclose(2, m.sv[3])) - - # def test_cy_equal_np_ft_random(self): - # TODO - # ft = FastText(size=20, min_count=1) - # ft.build_vocab(SENTENCES) - - # m1 = Average(ft) - # m1.prep.prepare_vectors( - # sv=m1.sv, total_sentences=len(self.sentences), update=False - # ) - # m1._pre_train_calls() - - # from fse.models.average_inner import MAX_NGRAMS_IN_BATCH - - # m1.batch_ngrams = MAX_NGRAMS_IN_BATCH - # mem1 = m1._get_thread_working_mem() - # o1 = train_average_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - - # m2 = Average(ft) - # m2.prep.prepare_vectors( - # sv=m2.sv, total_sentences=len(self.sentences), update=False - # ) - # m2._pre_train_calls() - # mem2 = m2._get_thread_working_mem() - - # from fse.models.average_inner import train_average_cy - - # o2 = train_average_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) - - # self.assertEqual(o1, o2) - # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - - # def test_train_single_from_disk(self): - # TODO - # p = Path("fse/test/test_data/test_vecs") - # p_res = Path("fse/test/test_data/test_vecs.vectors") - # p_target = Path("fse/test/test_data/test_vecs_wv.vectors") - - # se1 = Average(W2V) - # se2 = Average( - # W2V, sv_mapfile_path=str(p.absolute()), wv_mapfile_path=str(p.absolute()) - # ) - # se1.train([(s, i) for i, s in enumerate(SENTENCES)]) - # se2.train([(s, i) for i, s in enumerate(SENTENCES)]) - - # self.assertTrue(p_target.exists()) - # self.assertTrue((se1.wv.vectors == se2.wv.vectors).all()) - # self.assertFalse(se2.wv.vectors.flags.writeable) - - # self.assertTrue((se1.sv.vectors == se2.sv.vectors).all()) - # p_res.unlink() - # p_target.unlink() - - # def test_train_multi_from_disk(self): - # TODO - # p = Path("fse/test/test_data/test_vecs") - # p_res = Path("fse/test/test_data/test_vecs.vectors") - # p_target = Path("fse/test/test_data/test_vecs_wv.vectors") - - # se1 = Average(W2V, workers=2) - # se2 = Average( - # W2V, - # workers=2, - # sv_mapfile_path=str(p.absolute()), - # wv_mapfile_path=str(p.absolute()), - # ) - # se1.train([(s, i) for i, s in enumerate(SENTENCES)]) - # se2.train([(s, i) for i, s in enumerate(SENTENCES)]) - - # self.assertTrue(p_target.exists()) - # self.assertTrue((se1.wv.vectors == se2.wv.vectors).all()) - # self.assertFalse(se2.wv.vectors.flags.writeable) - - # self.assertTrue((se1.sv.vectors == se2.sv.vectors).all()) - # p_res.unlink() - # p_target.unlink() - def test_check_parameter_sanity(self): se = MaxPooling(W2V) se.word_weights = np.full(20, 2.0, dtype=np.float32) @@ -228,7 +131,7 @@ def test_pool_train_cy_w2v(self): self.assertTrue((306 == self.model.sv[1]).all()) self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) - def test_cy_equal_np_w2v(self): + def test_pool_cy_equal_np_w2v(self): m1 = MaxPooling(W2V) m1.prep.prepare_vectors( sv=m1.sv, total_sentences=len(self.sentences), update=False @@ -251,7 +154,7 @@ def test_cy_equal_np_w2v(self): self.assertEqual(o1, o2) self.assertTrue((m1.sv.vectors == m2.sv.vectors).all()) - def test_cy_equal_np_w2v_random(self): + def test_pool_cy_equal_np_w2v_random(self): w2v = Word2Vec(min_count=1, size=DIM) # Random initialization w2v.build_vocab(SENTENCES) @@ -276,6 +179,76 @@ def test_cy_equal_np_w2v_random(self): o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + def test_pool_train_np_ft(self): + m = MaxPooling(FT) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + mem = m._get_thread_working_mem() + + output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + self.assertTrue((241 == m.sv[0]).all()) + self.assertTrue( + np.allclose(737413.9, m.sv[2]) + ) + self.assertTrue( + np.allclose(1080970.2, m.sv[3]) + ) + + def test_pool_train_cy_ft(self): + m = MaxPooling(FT) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + mem = m._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + self.assertTrue((241 == m.sv[0]).all()) + self.assertTrue( + np.allclose(737413.9, m.sv[2]) + ) + self.assertTrue( + np.allclose(1080970.2, m.sv[3]) + ) + + def test_cy_equal_np_ft_random(self): + ft = FastText(size=20, min_count=1) + ft.build_vocab(SENTENCES) + + m1 = MaxPooling(ft) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + + from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH + + m1.batch_ngrams = MAX_NGRAMS_IN_BATCH + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) + + m2 = MaxPooling(ft) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) + + self.assertEqual(o1, o2) + self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) def test_pooling_train_np_w2v_non_negative(self): mpool = MaxPooling(W2V_R) @@ -302,21 +275,6 @@ def test_hpooling_train_np_w2v_non_negative(self): mpool.train(self.sentences) self.assertTrue((mpool.sv.vectors >= 0).all()) - def test_pooling_train_np_ft(self): - m = MaxPooling(FT) - m.prep.prepare_vectors( - sv=m.sv, total_sentences=len(self.sentences), update=False - ) - m._pre_train_calls() - mem = m._get_thread_working_mem() - - output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) - - self.assertEqual((5, 19), output) - self.assertTrue((241 == m.sv[0]).all()) - self.assertTrue((737413.9 == m.sv[2]).all()) - self.assertTrue((1080970.2 == m.sv[3]).all()) - def test_pooling_train_np_ft_non_negative(self): mpool = MaxPooling(FT_R) mpool.train(self.sentences) From 28ecf2ffba6d40f911015b5df17e9620b6973345 Mon Sep 17 00:00:00 2001 From: OliverB Date: Thu, 27 Feb 2020 10:05:08 +0100 Subject: [PATCH 22/63] Changed Date to 2020 --- README.md | 4 ++-- fse/inputs.py | 2 +- fse/models/average.py | 2 +- fse/models/average_inner.pxd | 2 +- fse/models/average_inner.pyx | 2 +- fse/models/base_s2v.py | 2 +- fse/models/pooling.py | 2 +- fse/models/pooling_inner.pyx | 2 +- fse/models/sentencevectors.py | 2 +- fse/models/sif.py | 2 +- fse/models/usif.py | 2 +- fse/models/utils.py | 2 +- fse/test/test_average.py | 2 +- fse/test/test_base_s2v.py | 2 +- fse/test/test_inputs.py | 2 +- fse/test/test_pooling.py | 2 +- fse/test/test_sentencevectors.py | 2 +- 17 files changed, 18 insertions(+), 18 deletions(-) diff --git a/README.md b/README.md index 8fcde07..05e8480 100644 --- a/README.md +++ b/README.md @@ -196,7 +196,7 @@ Copyright Author: Oliver Borchers -Copyright (C) 2019 Oliver Borchers +Copyright (C) 2020 Oliver Borchers Citation ------------- @@ -206,7 +206,7 @@ If you found this software useful, please cite it in your publication. @misc{Borchers2019, author = {Borchers, Oliver}, title = {Fast sentence embeddings}, - year = {2019}, + year = {2020}, publisher = {GitHub}, journal = {GitHub Repository}, howpublished = {\url{https://github.com/oborchers/Fast_Sentence_Embeddings}}, diff --git a/fse/inputs.py b/fse/inputs.py index d69d397..690c57d 100644 --- a/fse/inputs.py +++ b/fse/inputs.py @@ -2,7 +2,7 @@ # -*- coding: utf-8 -*- # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers from typing import MutableSequence diff --git a/fse/models/average.py b/fse/models/average.py index 02e2adc..617b5f6 100644 --- a/fse/models/average.py +++ b/fse/models/average.py @@ -2,7 +2,7 @@ # -*- coding: utf-8 -*- # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers """This module implements the base class to compute average representations for sentences, using highly optimized C routines, data streaming and Pythonic interfaces. diff --git a/fse/models/average_inner.pxd b/fse/models/average_inner.pxd index 1cdc509..66f56af 100644 --- a/fse/models/average_inner.pxd +++ b/fse/models/average_inner.pxd @@ -5,7 +5,7 @@ # coding: utf-8 # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers cimport numpy as np diff --git a/fse/models/average_inner.pyx b/fse/models/average_inner.pyx index 5694e2f..8ac29a6 100644 --- a/fse/models/average_inner.pyx +++ b/fse/models/average_inner.pyx @@ -6,7 +6,7 @@ # coding: utf-8 # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers """Optimized cython functions for computing sentence embeddings""" diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index 5fde303..177b31a 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -2,7 +2,7 @@ # -*- coding: utf-8 -*- # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers # Licensed under GNU General Public License v3.0 """Base class containing common methods for training, using & evaluating sentence embeddings. diff --git a/fse/models/pooling.py b/fse/models/pooling.py index 855f452..8fa9bb9 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -2,7 +2,7 @@ # -*- coding: utf-8 -*- # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers """This module implements the base class to compute Max Pooling representations for sentences, using highly optimized C routines, data streaming and Pythonic interfaces. diff --git a/fse/models/pooling_inner.pyx b/fse/models/pooling_inner.pyx index 2c7313b..4ce80b4 100644 --- a/fse/models/pooling_inner.pyx +++ b/fse/models/pooling_inner.pyx @@ -6,7 +6,7 @@ # coding: utf-8 # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers """Optimized cython functions for computing sentence embeddings""" diff --git a/fse/models/sentencevectors.py b/fse/models/sentencevectors.py index 8d2ddb0..44c4f28 100644 --- a/fse/models/sentencevectors.py +++ b/fse/models/sentencevectors.py @@ -2,7 +2,7 @@ # -*- coding: utf-8 -*- # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers from __future__ import division diff --git a/fse/models/sif.py b/fse/models/sif.py index 6b57b9f..edb7c7e 100644 --- a/fse/models/sif.py +++ b/fse/models/sif.py @@ -2,7 +2,7 @@ # -*- coding: utf-8 -*- # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers from fse.models.average import Average from fse.models.utils import compute_principal_components, remove_principal_components diff --git a/fse/models/usif.py b/fse/models/usif.py index caa3b68..341593a 100644 --- a/fse/models/usif.py +++ b/fse/models/usif.py @@ -2,7 +2,7 @@ # -*- coding: utf-8 -*- # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers from fse.models.average import Average from fse.models.utils import compute_principal_components, remove_principal_components diff --git a/fse/models/utils.py b/fse/models/utils.py index 190f97f..eb6bd39 100644 --- a/fse/models/utils.py +++ b/fse/models/utils.py @@ -2,7 +2,7 @@ # -*- coding: utf-8 -*- # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers from sklearn.decomposition import TruncatedSVD diff --git a/fse/test/test_average.py b/fse/test/test_average.py index 0bfd02f..13c9f38 100644 --- a/fse/test/test_average.py +++ b/fse/test/test_average.py @@ -2,7 +2,7 @@ # -*- coding: utf-8 -*- # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers """ Automated tests for checking the average model. diff --git a/fse/test/test_base_s2v.py b/fse/test/test_base_s2v.py index ed03687..192dc1e 100644 --- a/fse/test/test_base_s2v.py +++ b/fse/test/test_base_s2v.py @@ -2,7 +2,7 @@ # -*- coding: utf-8 -*- # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers """ Automated tests for checking the base_s2v class. diff --git a/fse/test/test_inputs.py b/fse/test/test_inputs.py index 7aec523..5304406 100644 --- a/fse/test/test_inputs.py +++ b/fse/test/test_inputs.py @@ -2,7 +2,7 @@ # -*- coding: utf-8 -*- # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers """ diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index e477851..ec28ed3 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -2,7 +2,7 @@ # -*- coding: utf-8 -*- # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers """ Automated tests for checking the model. diff --git a/fse/test/test_sentencevectors.py b/fse/test/test_sentencevectors.py index 1fb18ec..0314b6f 100644 --- a/fse/test/test_sentencevectors.py +++ b/fse/test/test_sentencevectors.py @@ -2,7 +2,7 @@ # -*- coding: utf-8 -*- # Author: Oliver Borchers -# Copyright (C) 2019 Oliver Borchers +# Copyright (C) 2020 Oliver Borchers """ From 947ebb26f528d05798fe8a5605774de2cbed9d05 Mon Sep 17 00:00:00 2001 From: OliverB Date: Thu, 27 Feb 2020 10:06:57 +0100 Subject: [PATCH 23/63] Added GPL_v3 --- fse/test/test_average.py | 1 + fse/test/test_base_s2v.py | 2 ++ fse/test/test_inputs.py | 1 + fse/test/test_pooling.py | 1 + fse/test/test_sentencevectors.py | 1 + fse/test/test_sif.py | 6 ++++++ fse/test/test_usif.py | 6 ++++++ fse/test/test_utils.py | 6 ++++++ 8 files changed, 24 insertions(+) diff --git a/fse/test/test_average.py b/fse/test/test_average.py index 13c9f38..27f066e 100644 --- a/fse/test/test_average.py +++ b/fse/test/test_average.py @@ -3,6 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 """ Automated tests for checking the average model. diff --git a/fse/test/test_base_s2v.py b/fse/test/test_base_s2v.py index 192dc1e..818965b 100644 --- a/fse/test/test_base_s2v.py +++ b/fse/test/test_base_s2v.py @@ -3,6 +3,8 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 + """ Automated tests for checking the base_s2v class. diff --git a/fse/test/test_inputs.py b/fse/test/test_inputs.py index 5304406..8c050e5 100644 --- a/fse/test/test_inputs.py +++ b/fse/test/test_inputs.py @@ -3,6 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 """ diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index ec28ed3..9cc35dc 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -3,6 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 """ Automated tests for checking the model. diff --git a/fse/test/test_sentencevectors.py b/fse/test/test_sentencevectors.py index 0314b6f..4a7680a 100644 --- a/fse/test/test_sentencevectors.py +++ b/fse/test/test_sentencevectors.py @@ -3,6 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 """ diff --git a/fse/test/test_sif.py b/fse/test/test_sif.py index d784b64..e44a704 100644 --- a/fse/test/test_sif.py +++ b/fse/test/test_sif.py @@ -1,3 +1,9 @@ +#!/usr/bin/env python +# -*- coding: utf-8 -*- + +# Author: Oliver Borchers +# Copyright (C) 2020 Oliver Borchers + import logging import unittest diff --git a/fse/test/test_usif.py b/fse/test/test_usif.py index fe5320f..b9ba90d 100644 --- a/fse/test/test_usif.py +++ b/fse/test/test_usif.py @@ -1,3 +1,9 @@ +#!/usr/bin/env python +# -*- coding: utf-8 -*- + +# Author: Oliver Borchers +# Copyright (C) 2020 Oliver Borchers + import logging import unittest diff --git a/fse/test/test_utils.py b/fse/test/test_utils.py index 8190c8b..550e540 100644 --- a/fse/test/test_utils.py +++ b/fse/test/test_utils.py @@ -1,3 +1,9 @@ +#!/usr/bin/env python +# -*- coding: utf-8 -*- + +# Author: Oliver Borchers +# Copyright (C) 2020 Oliver Borchers + import logging import unittest From 8a020bd3223dac71e903dc6de5bf358babcd77aa Mon Sep 17 00:00:00 2001 From: OliverB Date: Thu, 27 Feb 2020 10:08:25 +0100 Subject: [PATCH 24/63] Added GPL v3 --- fse/models/average.py | 1 + fse/models/average_inner.pxd | 1 + fse/models/average_inner.pyx | 1 + fse/models/pooling.py | 1 + fse/models/pooling_inner.pyx | 1 + fse/models/sentencevectors.py | 1 + fse/models/sif.py | 1 + fse/models/usif.py | 1 + fse/models/utils.py | 1 + 9 files changed, 9 insertions(+) diff --git a/fse/models/average.py b/fse/models/average.py index 617b5f6..b068dde 100644 --- a/fse/models/average.py +++ b/fse/models/average.py @@ -3,6 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 """This module implements the base class to compute average representations for sentences, using highly optimized C routines, data streaming and Pythonic interfaces. diff --git a/fse/models/average_inner.pxd b/fse/models/average_inner.pxd index 66f56af..ed245ae 100644 --- a/fse/models/average_inner.pxd +++ b/fse/models/average_inner.pxd @@ -6,6 +6,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 cimport numpy as np diff --git a/fse/models/average_inner.pyx b/fse/models/average_inner.pyx index 8ac29a6..f301b93 100644 --- a/fse/models/average_inner.pyx +++ b/fse/models/average_inner.pyx @@ -7,6 +7,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 """Optimized cython functions for computing sentence embeddings""" diff --git a/fse/models/pooling.py b/fse/models/pooling.py index 8fa9bb9..297f7f3 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -3,6 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 """This module implements the base class to compute Max Pooling representations for sentences, using highly optimized C routines, data streaming and Pythonic interfaces. diff --git a/fse/models/pooling_inner.pyx b/fse/models/pooling_inner.pyx index 4ce80b4..1f51fdf 100644 --- a/fse/models/pooling_inner.pyx +++ b/fse/models/pooling_inner.pyx @@ -7,6 +7,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 """Optimized cython functions for computing sentence embeddings""" diff --git a/fse/models/sentencevectors.py b/fse/models/sentencevectors.py index 44c4f28..6a3ac88 100644 --- a/fse/models/sentencevectors.py +++ b/fse/models/sentencevectors.py @@ -3,6 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 from __future__ import division diff --git a/fse/models/sif.py b/fse/models/sif.py index edb7c7e..6eeca2f 100644 --- a/fse/models/sif.py +++ b/fse/models/sif.py @@ -3,6 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 from fse.models.average import Average from fse.models.utils import compute_principal_components, remove_principal_components diff --git a/fse/models/usif.py b/fse/models/usif.py index 341593a..29d6321 100644 --- a/fse/models/usif.py +++ b/fse/models/usif.py @@ -3,6 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 from fse.models.average import Average from fse.models.utils import compute_principal_components, remove_principal_components diff --git a/fse/models/utils.py b/fse/models/utils.py index eb6bd39..aebab8a 100644 --- a/fse/models/utils.py +++ b/fse/models/utils.py @@ -3,6 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers +# Licensed under GNU General Public License v3.0 from sklearn.decomposition import TruncatedSVD From 0f848cba3dbdae8c34aa8d7cbaf2e0cd3ed7db81 Mon Sep 17 00:00:00 2001 From: OliverB Date: Thu, 27 Feb 2020 10:09:38 +0100 Subject: [PATCH 25/63] Changed setup_requires --- setup.py | 4 ++++ 1 file changed, 4 insertions(+) diff --git a/setup.py b/setup.py index f305b20..e904c95 100755 --- a/setup.py +++ b/setup.py @@ -107,6 +107,10 @@ def finalize_options(self): test_suite="fse.test", + setup_requires=[ + 'numpy >= 1.11.3', + ], + install_requires=[ 'numpy >= 1.11.3', 'scipy >= 0.18.1', From 902822fab5fe49654f91269988940b80ada4cd81 Mon Sep 17 00:00:00 2001 From: OliverB Date: Thu, 27 Feb 2020 10:10:00 +0100 Subject: [PATCH 26/63] black --- setup.py | 93 +++++++++++++++++++++++++++----------------------------- 1 file changed, 45 insertions(+), 48 deletions(-) diff --git a/setup.py b/setup.py index e904c95..8c14e25 100755 --- a/setup.py +++ b/setup.py @@ -1,11 +1,11 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -''' +""" Run with: sudo python ./setup.py install -''' +""" import os import sys @@ -14,12 +14,14 @@ from setuptools.command.build_ext import build_ext if sys.version_info[:2] < (3, 6): - raise Exception('This version of fse needs Python 3.6 or later.') + raise Exception("This version of fse needs Python 3.6 or later.") + class custom_build_ext(build_ext): - '''Allow C extension building to fail. - ''' - warning_message = ''' + """Allow C extension building to fail. + """ + + warning_message = """ ******************************************************************** WARNING: %s could not be compiled. %s @@ -42,18 +44,19 @@ class custom_build_ext(build_ext): http://api.mongodb.org/python/current/installation.html#osx ******************************************************************** -''' +""" def run(self): try: build_ext.run(self) except Exception: e = sys.exc_info()[1] - sys.stdout.write('%s\n' % str(e)) + sys.stdout.write("%s\n" % str(e)) warnings.warn( - self.warning_message + - 'Extension modules' + - 'There was an issue with your platform configuration - see above.') + self.warning_message + + "Extension modules" + + "There was an issue with your platform configuration - see above." + ) def build_extension(self, ext): name = ext.name @@ -61,64 +64,58 @@ def build_extension(self, ext): build_ext.build_extension(self, ext) except Exception: e = sys.exc_info()[1] - sys.stdout.write('%s\n' % str(e)) + sys.stdout.write("%s\n" % str(e)) warnings.warn( - self.warning_message + - 'The %s extension module' % (name,) + - 'The output above this warning shows how the compilation failed.') + self.warning_message + + "The %s extension module" % (name,) + + "The output above this warning shows how the compilation failed." + ) def finalize_options(self): build_ext.finalize_options(self) if isinstance(__builtins__, dict): - __builtins__['__NUMPY_SETUP__'] = False + __builtins__["__NUMPY_SETUP__"] = False else: __builtins__.__NUMPY_SETUP__ = False import numpy + self.include_dirs.append(numpy.get_include()) -mod_dir = os.path.join(os.path.dirname(__file__), 'fse', 'models') -fse_dir = os.path.join(os.path.dirname(__file__), 'fse') -cmdclass = {'build_ext': custom_build_ext} +mod_dir = os.path.join(os.path.dirname(__file__), "fse", "models") +fse_dir = os.path.join(os.path.dirname(__file__), "fse") -setup( - name='fse', - version='0.1.16', - description='Fast Sentence Embeddings for Gensim', - - author=u'Oliver Borchers', - author_email='borchers@bwl.uni-mannheim.de', +cmdclass = {"build_ext": custom_build_ext} +setup( + name="fse", + version="0.1.16", + description="Fast Sentence Embeddings for Gensim", + author=u"Oliver Borchers", + author_email="borchers@bwl.uni-mannheim.de", url="https://github.com/oborchers/Fast_Sentence_Embeddings", - - license='GPL-3.0', - + license="GPL-3.0", ext_modules=[ - Extension('fse.models.average_inner', - sources=['./fse/models/average_inner.pyx'], #.c - include_dirs=[mod_dir]), - ], - + Extension( + "fse.models.average_inner", + sources=["./fse/models/average_inner.pyx"], # .c + include_dirs=[mod_dir], + ), + ], cmdclass=cmdclass, packages=find_packages(), - zip_safe=False, - test_suite="fse.test", - - setup_requires=[ - 'numpy >= 1.11.3', - ], - + setup_requires=["numpy >= 1.11.3",], install_requires=[ - 'numpy >= 1.11.3', - 'scipy >= 0.18.1', - 'smart_open >= 1.5.0', - 'scikit-learn >= 0.19.1', - 'gensim >= 3.8.0', - 'wordfreq >= 2.2.1', - 'psutil' + "numpy >= 1.11.3", + "scipy >= 0.18.1", + "smart_open >= 1.5.0", + "scikit-learn >= 0.19.1", + "gensim >= 3.8.0", + "wordfreq >= 2.2.1", + "psutil", ], include_package_data=True, ) From 66a900122f9626cb8d3d285cfc0d6820c5a9538a Mon Sep 17 00:00:00 2001 From: OliverB Date: Thu, 27 Feb 2020 10:13:55 +0100 Subject: [PATCH 27/63] Changed setup --- MANIFEST.in | 4 +++- setup.py | 5 +++++ 2 files changed, 8 insertions(+), 1 deletion(-) diff --git a/MANIFEST.in b/MANIFEST.in index 5b2ce84..b739009 100644 --- a/MANIFEST.in +++ b/MANIFEST.in @@ -4,4 +4,6 @@ include README.md include fse/models/voidptr.h include fse/models/average_inner.pyx -include fse/models/average_inner.pxd \ No newline at end of file +include fse/models/average_inner.pxd + +include fse/models/pooling_inner.pyx \ No newline at end of file diff --git a/setup.py b/setup.py index 8c14e25..2a121c2 100755 --- a/setup.py +++ b/setup.py @@ -102,6 +102,11 @@ def finalize_options(self): sources=["./fse/models/average_inner.pyx"], # .c include_dirs=[mod_dir], ), + Extension( + "fse.models.pooling_inner", + sources=["./fse/models/pooling_inner.pyx"], # .c + include_dirs=[mod_dir], + ), ], cmdclass=cmdclass, packages=find_packages(), From f8d664f9cc3c3b36782e53b81b3b6a1bac086c7c Mon Sep 17 00:00:00 2001 From: OliverB Date: Thu, 27 Feb 2020 18:51:08 +0100 Subject: [PATCH 28/63] Reformat --- fse/models/pooling_inner.c | 740 ++++++++++++++++++++++------------- fse/models/pooling_inner.pyx | 92 ++++- 2 files changed, 546 insertions(+), 286 deletions(-) diff --git a/fse/models/pooling_inner.c b/fse/models/pooling_inner.c index d8a72d0..8164dc1 100644 --- a/fse/models/pooling_inner.c +++ b/fse/models/pooling_inner.c @@ -1056,7 +1056,7 @@ typedef npy_double __pyx_t_5numpy_double_t; */ typedef npy_longdouble __pyx_t_5numpy_longdouble_t; -/* "average_inner.pxd":15 +/* "average_inner.pxd":16 * void* PyCObject_AsVoidPtr(object obj) * * ctypedef np.float32_t REAL_t # <<<<<<<<<<<<<< @@ -1065,7 +1065,7 @@ typedef npy_longdouble __pyx_t_5numpy_longdouble_t; */ typedef __pyx_t_5numpy_float32_t __pyx_t_13average_inner_REAL_t; -/* "average_inner.pxd":16 +/* "average_inner.pxd":17 * * ctypedef np.float32_t REAL_t * ctypedef np.uint32_t uINT_t # <<<<<<<<<<<<<< @@ -1138,7 +1138,7 @@ typedef npy_cdouble __pyx_t_5numpy_complex_t; struct __pyx_t_13average_inner_BaseSentenceVecsConfig; struct __pyx_t_13average_inner_FTSentenceVecsConfig; -/* "average_inner.pxd":19 +/* "average_inner.pxd":20 * * # BLAS routine signatures * ctypedef void (*saxpy_ptr) (const int *N, const float *alpha, const float *X, const int *incX, float *Y, const int *incY) nogil # <<<<<<<<<<<<<< @@ -1147,7 +1147,7 @@ struct __pyx_t_13average_inner_FTSentenceVecsConfig; */ typedef void (*__pyx_t_13average_inner_saxpy_ptr)(int const *, float const *, float const *, int const *, float *, int const *); -/* "average_inner.pxd":20 +/* "average_inner.pxd":21 * # BLAS routine signatures * ctypedef void (*saxpy_ptr) (const int *N, const float *alpha, const float *X, const int *incX, float *Y, const int *incY) nogil * ctypedef void (*sscal_ptr) (const int *N, const float *alpha, const float *X, const int *incX) nogil # <<<<<<<<<<<<<< @@ -1156,7 +1156,7 @@ typedef void (*__pyx_t_13average_inner_saxpy_ptr)(int const *, float const *, fl */ typedef void (*__pyx_t_13average_inner_sscal_ptr)(int const *, float const *, float const *, int const *); -/* "average_inner.pxd":34 +/* "average_inner.pxd":35 * DEF MAX_NGRAMS = 40 * * cdef struct BaseSentenceVecsConfig: # <<<<<<<<<<<<<< @@ -1176,7 +1176,7 @@ struct __pyx_t_13average_inner_BaseSentenceVecsConfig { __pyx_t_13average_inner_uINT_t sentence_boundary[(0x2710 + 1)]; }; -/* "average_inner.pxd":48 +/* "average_inner.pxd":49 * uINT_t sentence_boundary[MAX_WORDS + 1] * * cdef struct FTSentenceVecsConfig: # <<<<<<<<<<<<<< @@ -1725,6 +1725,7 @@ static PyObject *(*__pyx_f_13average_inner_populate_ft_s2v_config)(struct __pyx_ /* Module declarations from 'pooling_inner' */ static void __pyx_f_13pooling_inner_sl_max_pool(int const *, float *, float const *); /*proto*/ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ +static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_uINT_t); /*proto*/ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ #define __Pyx_MODULE_NAME "pooling_inner" extern int __pyx_module_is_main_pooling_inner; @@ -1756,6 +1757,7 @@ static const char __pyx_k_target[] = "target"; static const char __pyx_k_eff_words[] = "eff_words"; static const char __pyx_k_ValueError[] = "ValueError"; static const char __pyx_k_ImportError[] = "ImportError"; +static const char __pyx_k_window_size[] = "window_size"; static const char __pyx_k_FAST_VERSION[] = "FAST_VERSION"; static const char __pyx_k_RuntimeError[] = "RuntimeError"; static const char __pyx_k_hierarchical[] = "hierarchical"; @@ -1817,6 +1819,7 @@ static PyObject *__pyx_n_s_train_pooling_cy; static PyObject *__pyx_kp_u_unknown_dtype_code_in_numpy_pxd; static PyObject *__pyx_n_s_vocab; static PyObject *__pyx_n_s_w2v; +static PyObject *__pyx_n_s_window_size; static PyObject *__pyx_n_s_wv; static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObject *__pyx_self, PyObject *__pyx_v_model, PyObject *__pyx_v_indexed_sentences, PyObject *__pyx_v_target, PyObject *__pyx_v_memory); /* proto */ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_self); /* proto */ @@ -1837,12 +1840,12 @@ static PyObject *__pyx_codeobj__10; static PyObject *__pyx_codeobj__11; /* Late includes */ -/* "pooling_inner.pyx":42 +/* "pooling_inner.pyx":43 * DEF MAX_NGRAMS = 40 * - * cdef void sl_max_pool(const int *N, float *X, const float *Y) nogil: # <<<<<<<<<<<<<< - * """ Performs single left max pooling op - * + * cdef void sl_max_pool( # <<<<<<<<<<<<<< + * const int *N, + * float *X, */ static void __pyx_f_13pooling_inner_sl_max_pool(int const *__pyx_v_N, float *__pyx_v_X, float const *__pyx_v_Y) { @@ -1850,7 +1853,7 @@ static void __pyx_f_13pooling_inner_sl_max_pool(int const *__pyx_v_N, float *__p int __pyx_t_1; int __pyx_t_2; - /* "pooling_inner.pyx":56 + /* "pooling_inner.pyx":61 * """ * cdef int i * for i from 0 <= i < N[0] by 1: # <<<<<<<<<<<<<< @@ -1860,7 +1863,7 @@ static void __pyx_f_13pooling_inner_sl_max_pool(int const *__pyx_v_N, float *__p __pyx_t_1 = (__pyx_v_N[0]); for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_1; __pyx_v_i+=1) { - /* "pooling_inner.pyx":57 + /* "pooling_inner.pyx":62 * cdef int i * for i from 0 <= i < N[0] by 1: * if X[i] < Y[i]: # <<<<<<<<<<<<<< @@ -1870,16 +1873,16 @@ static void __pyx_f_13pooling_inner_sl_max_pool(int const *__pyx_v_N, float *__p __pyx_t_2 = (((__pyx_v_X[__pyx_v_i]) < (__pyx_v_Y[__pyx_v_i])) != 0); if (__pyx_t_2) { - /* "pooling_inner.pyx":58 + /* "pooling_inner.pyx":63 * for i from 0 <= i < N[0] by 1: * if X[i] < Y[i]: * X[i] = Y[i] # <<<<<<<<<<<<<< * - * cdef void compute_base_sentence_pooling(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: + * cdef void compute_base_sentence_pooling( */ (__pyx_v_X[__pyx_v_i]) = (__pyx_v_Y[__pyx_v_i]); - /* "pooling_inner.pyx":57 + /* "pooling_inner.pyx":62 * cdef int i * for i from 0 <= i < N[0] by 1: * if X[i] < Y[i]: # <<<<<<<<<<<<<< @@ -1889,23 +1892,23 @@ static void __pyx_f_13pooling_inner_sl_max_pool(int const *__pyx_v_N, float *__p } } - /* "pooling_inner.pyx":42 + /* "pooling_inner.pyx":43 * DEF MAX_NGRAMS = 40 * - * cdef void sl_max_pool(const int *N, float *X, const float *Y) nogil: # <<<<<<<<<<<<<< - * """ Performs single left max pooling op - * + * cdef void sl_max_pool( # <<<<<<<<<<<<<< + * const int *N, + * float *X, */ /* function exit code */ } -/* "pooling_inner.pyx":60 +/* "pooling_inner.pyx":65 * X[i] = Y[i] * - * cdef void compute_base_sentence_pooling(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< - * """Perform optimized sentence-level max pooling for BaseAny2Vec model. - * + * cdef void compute_base_sentence_pooling( # <<<<<<<<<<<<<< + * BaseSentenceVecsConfig *c, + * uINT_t num_sentences, */ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { @@ -1925,7 +1928,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t __pyx_t_13average_inner_uINT_t __pyx_t_6; __pyx_t_13average_inner_uINT_t __pyx_t_7; - /* "pooling_inner.pyx":76 + /* "pooling_inner.pyx":84 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -1935,47 +1938,29 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":84 + /* "pooling_inner.pyx":92 * REAL_t sent_len, inv_count * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< - * memset(c.mem, 0, size * cython.sizeof(REAL_t)) - * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] */ __pyx_t_2 = __pyx_v_num_sentences; __pyx_t_3 = __pyx_t_2; for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) { __pyx_v_sent_idx = __pyx_t_4; - /* "pooling_inner.pyx":85 + /* "pooling_inner.pyx":93 * * for sent_idx in range(num_sentences): - * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< - * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) - * - */ - (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - - /* "pooling_inner.pyx":86 - * for sent_idx in range(num_sentences): - * memset(c.mem, 0, size * cython.sizeof(REAL_t)) - * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< - * - * sent_start = c.sentence_boundary[sent_idx] - */ - (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - - /* "pooling_inner.pyx":88 - * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) - * * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":89 - * + /* "pooling_inner.pyx":94 + * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< * sent_len = ZEROF @@ -1983,7 +1968,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":90 + /* "pooling_inner.pyx":95 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -1992,7 +1977,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":92 + /* "pooling_inner.pyx":97 * sent_len = ZEROF * * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -2004,7 +1989,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { __pyx_v_i = __pyx_t_7; - /* "pooling_inner.pyx":93 + /* "pooling_inner.pyx":98 * * for i in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -2013,7 +1998,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":94 + /* "pooling_inner.pyx":99 * for i in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< @@ -2022,7 +2007,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); - /* "pooling_inner.pyx":95 + /* "pooling_inner.pyx":100 * sent_len += ONEF * sent_row = c.sent_adresses[i] * size * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< @@ -2031,7 +2016,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); - /* "pooling_inner.pyx":97 + /* "pooling_inner.pyx":102 * word_row = c.word_indices[i] * size * * sl_max_pool( # <<<<<<<<<<<<<< @@ -2042,23 +2027,149 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t } } - /* "pooling_inner.pyx":60 + /* "pooling_inner.pyx":65 * X[i] = Y[i] * - * cdef void compute_base_sentence_pooling(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< - * """Perform optimized sentence-level max pooling for BaseAny2Vec model. - * + * cdef void compute_base_sentence_pooling( # <<<<<<<<<<<<<< + * BaseSentenceVecsConfig *c, + * uINT_t num_sentences, */ /* function exit code */ } -/* "pooling_inner.pyx":105 +/* "pooling_inner.pyx":110 + * + * + * cdef void compute_base_sentence_hier_pooling( # <<<<<<<<<<<<<< + * BaseSentenceVecsConfig *c, + * uINT_t num_sentences, + */ + +static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences, CYTHON_UNUSED __pyx_t_13average_inner_uINT_t __pyx_v_window_size) { + int __pyx_v_size; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_end; + CYTHON_UNUSED __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; + __pyx_t_13average_inner_uINT_t __pyx_v_i; + CYTHON_UNUSED __pyx_t_13average_inner_uINT_t __pyx_v_word_row; + __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; + int __pyx_t_1; + __pyx_t_13average_inner_uINT_t __pyx_t_2; + __pyx_t_13average_inner_uINT_t __pyx_t_3; + __pyx_t_13average_inner_uINT_t __pyx_t_4; + __pyx_t_13average_inner_uINT_t __pyx_t_5; + __pyx_t_13average_inner_uINT_t __pyx_t_6; + __pyx_t_13average_inner_uINT_t __pyx_t_7; + + /* "pooling_inner.pyx":130 + * """ + * cdef: + * int size = c.size # <<<<<<<<<<<<<< + * + * uINT_t sent_idx, sent_start, sent_end, sent_row + */ + __pyx_t_1 = __pyx_v_c->size; + __pyx_v_size = __pyx_t_1; + + /* "pooling_inner.pyx":138 + * REAL_t sent_len, inv_count + * + * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] + */ + __pyx_t_2 = __pyx_v_num_sentences; + __pyx_t_3 = __pyx_t_2; + for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) { + __pyx_v_sent_idx = __pyx_t_4; + + /* "pooling_inner.pyx":139 + * + * for sent_idx in range(num_sentences): + * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< + * sent_end = c.sentence_boundary[sent_idx + 1] + * sent_len = ZEROF + */ + __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); + + /* "pooling_inner.pyx":140 + * for sent_idx in range(num_sentences): + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< + * sent_len = ZEROF + * + */ + __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); + + /* "pooling_inner.pyx":141 + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] + * sent_len = ZEROF # <<<<<<<<<<<<<< + * + * for i in range(sent_start, sent_end): + */ + __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; + + /* "pooling_inner.pyx":143 + * sent_len = ZEROF + * + * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< + * sent_len += ONEF + * sent_row = c.sent_adresses[i] * size + */ + __pyx_t_5 = __pyx_v_sent_end; + __pyx_t_6 = __pyx_t_5; + for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { + __pyx_v_i = __pyx_t_7; + + /* "pooling_inner.pyx":144 + * + * for i in range(sent_start, sent_end): + * sent_len += ONEF # <<<<<<<<<<<<<< + * sent_row = c.sent_adresses[i] * size + * word_row = c.word_indices[i] * size + */ + __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); + + /* "pooling_inner.pyx":145 + * for i in range(sent_start, sent_end): + * sent_len += ONEF + * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< + * word_row = c.word_indices[i] * size + * + */ + __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); + + /* "pooling_inner.pyx":146 + * sent_len += ONEF + * sent_row = c.sent_adresses[i] * size + * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< + * + * + */ + __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); + } + } + + /* "pooling_inner.pyx":110 * * - * cdef void compute_ft_sentence_pooling(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< - * """Perform optimized sentence-level max pooling for FastText model. + * cdef void compute_base_sentence_hier_pooling( # <<<<<<<<<<<<<< + * BaseSentenceVecsConfig *c, + * uINT_t num_sentences, + */ + + /* function exit code */ +} + +/* "pooling_inner.pyx":149 + * * + * cdef void compute_ft_sentence_pooling( # <<<<<<<<<<<<<< + * FTSentenceVecsConfig *c, + * uINT_t num_sentences, */ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { @@ -2089,7 +2200,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_13average_inner_uINT_t __pyx_t_11; __pyx_t_13average_inner_uINT_t __pyx_t_12; - /* "pooling_inner.pyx":121 + /* "pooling_inner.pyx":168 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -2099,7 +2210,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":131 + /* "pooling_inner.pyx":178 * REAL_t sent_len * REAL_t inv_count, inv_ngram * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< @@ -2109,7 +2220,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_2 = __pyx_v_c->oov_weight; __pyx_v_oov_weight = __pyx_t_2; - /* "pooling_inner.pyx":133 + /* "pooling_inner.pyx":180 * REAL_t oov_weight = c.oov_weight * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -2121,7 +2232,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { __pyx_v_sent_idx = __pyx_t_5; - /* "pooling_inner.pyx":134 + /* "pooling_inner.pyx":181 * * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -2130,7 +2241,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":135 + /* "pooling_inner.pyx":182 * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -2139,7 +2250,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":136 + /* "pooling_inner.pyx":183 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -2148,7 +2259,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":138 + /* "pooling_inner.pyx":185 * sent_len = ZEROF * * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -2160,7 +2271,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { __pyx_v_i = __pyx_t_8; - /* "pooling_inner.pyx":139 + /* "pooling_inner.pyx":186 * * for i in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -2169,7 +2280,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":140 + /* "pooling_inner.pyx":187 * for i in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< @@ -2178,7 +2289,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); - /* "pooling_inner.pyx":142 + /* "pooling_inner.pyx":189 * sent_row = c.sent_adresses[i] * size * * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< @@ -2187,7 +2298,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); - /* "pooling_inner.pyx":143 + /* "pooling_inner.pyx":190 * * word_idx = c.word_indices[i] * ngrams = c.subwords_idx_len[i] # <<<<<<<<<<<<<< @@ -2196,7 +2307,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_i]); - /* "pooling_inner.pyx":145 + /* "pooling_inner.pyx":192 * ngrams = c.subwords_idx_len[i] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2206,7 +2317,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":146 + /* "pooling_inner.pyx":193 * * if ngrams == 0: * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< @@ -2215,7 +2326,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); - /* "pooling_inner.pyx":148 + /* "pooling_inner.pyx":195 * word_row = c.word_indices[i] * size * * sl_max_pool( # <<<<<<<<<<<<<< @@ -2224,7 +2335,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_f_13pooling_inner_sl_max_pool((&__pyx_v_size), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row]))); - /* "pooling_inner.pyx":145 + /* "pooling_inner.pyx":192 * ngrams = c.subwords_idx_len[i] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2234,7 +2345,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 goto __pyx_L7; } - /* "pooling_inner.pyx":155 + /* "pooling_inner.pyx":202 * * else: * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2244,7 +2355,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 /*else*/ { (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":156 + /* "pooling_inner.pyx":203 * else: * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< @@ -2253,7 +2364,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); - /* "pooling_inner.pyx":157 + /* "pooling_inner.pyx":204 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight * for j in range(ngrams): # <<<<<<<<<<<<<< @@ -2265,7 +2376,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 for (__pyx_t_12 = 0; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { __pyx_v_j = __pyx_t_12; - /* "pooling_inner.pyx":158 + /* "pooling_inner.pyx":205 * inv_ngram = (ONEF / ngrams) * c.oov_weight * for j in range(ngrams): * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size # <<<<<<<<<<<<<< @@ -2274,7 +2385,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_i * 40) + __pyx_v_j)]) * __pyx_v_size); - /* "pooling_inner.pyx":159 + /* "pooling_inner.pyx":206 * for j in range(ngrams): * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size * saxpy( # <<<<<<<<<<<<<< @@ -2284,7 +2395,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_ngram), (&(__pyx_v_c->ngram_vectors[__pyx_v_ngram_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); } - /* "pooling_inner.pyx":168 + /* "pooling_inner.pyx":215 * ) * * sl_max_pool( # <<<<<<<<<<<<<< @@ -2297,23 +2408,23 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 } } - /* "pooling_inner.pyx":105 - * + /* "pooling_inner.pyx":149 * - * cdef void compute_ft_sentence_pooling(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< - * """Perform optimized sentence-level max pooling for FastText model. * + * cdef void compute_ft_sentence_pooling( # <<<<<<<<<<<<<< + * FTSentenceVecsConfig *c, + * uINT_t num_sentences, */ /* function exit code */ } -/* "pooling_inner.pyx":176 +/* "pooling_inner.pyx":223 * * - * def train_pooling_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< - * """Training on a sequence of sentences and update the target ndarray. - * + * def train_pooling_cy( # <<<<<<<<<<<<<< + * model, + * indexed_sentences, */ /* Python wrapper */ @@ -2355,23 +2466,23 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self case 1: if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 176, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 223, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 2: if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 176, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 223, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 3: if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 176, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 223, __pyx_L3_error) } } if (unlikely(kw_args > 0)) { - if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 176, __pyx_L3_error) + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 223, __pyx_L3_error) } } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { goto __pyx_L5_argtuple_error; @@ -2388,7 +2499,7 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self } goto __pyx_L4_argument_unpacking_done; __pyx_L5_argtuple_error:; - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 176, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 223, __pyx_L3_error) __pyx_L3_error:; __Pyx_AddTraceback("pooling_inner.train_pooling_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __Pyx_RefNannyFinishContext(); @@ -2404,152 +2515,174 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObject *__pyx_self, PyObject *__pyx_v_model, PyObject *__pyx_v_indexed_sentences, PyObject *__pyx_v_target, PyObject *__pyx_v_memory) { __pyx_t_13average_inner_uINT_t __pyx_v_eff_sentences; __pyx_t_13average_inner_uINT_t __pyx_v_eff_words; + __pyx_t_13average_inner_uINT_t __pyx_v_window_size; struct __pyx_t_13average_inner_BaseSentenceVecsConfig __pyx_v_w2v; struct __pyx_t_13average_inner_FTSentenceVecsConfig __pyx_v_ft; PyObject *__pyx_r = NULL; __Pyx_RefNannyDeclarations PyObject *__pyx_t_1 = NULL; - int __pyx_t_2; + __pyx_t_13average_inner_uINT_t __pyx_t_2; int __pyx_t_3; - PyObject *__pyx_t_4 = NULL; + int __pyx_t_4; PyObject *__pyx_t_5 = NULL; PyObject *__pyx_t_6 = NULL; - PyObject *(*__pyx_t_7)(PyObject *); - __pyx_t_13average_inner_uINT_t __pyx_t_8; + PyObject *__pyx_t_7 = NULL; + PyObject *(*__pyx_t_8)(PyObject *); __pyx_t_13average_inner_uINT_t __pyx_t_9; __Pyx_RefNannySetupContext("train_pooling_cy", 0); - /* "pooling_inner.pyx":200 + /* "pooling_inner.pyx":252 * """ * * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< * cdef uINT_t eff_words = 0 - * cdef BaseSentenceVecsConfig w2v + * cdef uINT_t window_size = model.window_size */ __pyx_v_eff_sentences = 0; - /* "pooling_inner.pyx":201 + /* "pooling_inner.pyx":253 * * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< + * cdef uINT_t window_size = model.window_size * cdef BaseSentenceVecsConfig w2v - * cdef FTSentenceVecsConfig ft */ __pyx_v_eff_words = 0; - /* "pooling_inner.pyx":205 + /* "pooling_inner.pyx":254 + * cdef uINT_t eff_sentences = 0 + * cdef uINT_t eff_words = 0 + * cdef uINT_t window_size = model.window_size # <<<<<<<<<<<<<< + * cdef BaseSentenceVecsConfig w2v + * cdef FTSentenceVecsConfig ft + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_size); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 254, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_1); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 254, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_v_window_size = ((__pyx_t_13average_inner_uINT_t)__pyx_t_2); + + /* "pooling_inner.pyx":258 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< * init_base_s2v_config(&w2v, model, target, memory) * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 205, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 258, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 205, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 258, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_3 = ((!__pyx_t_2) != 0); - if (__pyx_t_3) { + __pyx_t_4 = ((!__pyx_t_3) != 0); + if (__pyx_t_4) { - /* "pooling_inner.pyx":206 + /* "pooling_inner.pyx":259 * * if not model.is_ft: * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< * - * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + * eff_sentences, eff_words = populate_base_s2v_config( */ - __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 206, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 259, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":208 - * init_base_s2v_config(&w2v, model, target, memory) - * - * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) # <<<<<<<<<<<<<< - * - * if not model.hierarchical: + /* "pooling_inner.pyx":263 + * eff_sentences, eff_words = populate_base_s2v_config( + * &w2v, + * model.wv.vocab, # <<<<<<<<<<<<<< + * indexed_sentences + * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 208, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 263, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 208, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_4); + __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 263, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_5); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 208, __pyx_L1_error) + + /* "pooling_inner.pyx":261 + * init_base_s2v_config(&w2v, model, target, memory) + * + * eff_sentences, eff_words = populate_base_s2v_config( # <<<<<<<<<<<<<< + * &w2v, + * model.wv.vocab, + */ + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 261, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; + __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { PyObject* sequence = __pyx_t_1; Py_ssize_t size = __Pyx_PySequence_SIZE(sequence); if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 208, __pyx_L1_error) + __PYX_ERR(0, 261, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { - __pyx_t_4 = PyTuple_GET_ITEM(sequence, 0); - __pyx_t_5 = PyTuple_GET_ITEM(sequence, 1); + __pyx_t_5 = PyTuple_GET_ITEM(sequence, 0); + __pyx_t_6 = PyTuple_GET_ITEM(sequence, 1); } else { - __pyx_t_4 = PyList_GET_ITEM(sequence, 0); - __pyx_t_5 = PyList_GET_ITEM(sequence, 1); + __pyx_t_5 = PyList_GET_ITEM(sequence, 0); + __pyx_t_6 = PyList_GET_ITEM(sequence, 1); } - __Pyx_INCREF(__pyx_t_4); __Pyx_INCREF(__pyx_t_5); + __Pyx_INCREF(__pyx_t_6); #else - __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 208, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 208, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 261, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); + __pyx_t_6 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 261, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_6); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 208, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_6); + __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 261, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_7); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; - index = 0; __pyx_t_4 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_4)) goto __pyx_L4_unpacking_failed; - __Pyx_GOTREF(__pyx_t_4); - index = 1; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L4_unpacking_failed; + __pyx_t_8 = Py_TYPE(__pyx_t_7)->tp_iternext; + index = 0; __pyx_t_5 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_5)) goto __pyx_L4_unpacking_failed; __Pyx_GOTREF(__pyx_t_5); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 208, __pyx_L1_error) - __pyx_t_7 = NULL; - __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; + index = 1; __pyx_t_6 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_6)) goto __pyx_L4_unpacking_failed; + __Pyx_GOTREF(__pyx_t_6); + if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 261, __pyx_L1_error) + __pyx_t_8 = NULL; + __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; goto __pyx_L5_unpacking_done; __pyx_L4_unpacking_failed:; - __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; - __pyx_t_7 = NULL; + __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; + __pyx_t_8 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 208, __pyx_L1_error) + __PYX_ERR(0, 261, __pyx_L1_error) __pyx_L5_unpacking_done:; } - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 208, __pyx_L1_error) - __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 208, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 261, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; - __pyx_v_eff_sentences = __pyx_t_8; + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 261, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; + __pyx_v_eff_sentences = __pyx_t_2; __pyx_v_eff_words = __pyx_t_9; - /* "pooling_inner.pyx":210 - * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + /* "pooling_inner.pyx":267 + * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: - * compute_base_sentence_pooling(&w2v, eff_sentences) + * compute_base_sentence_pooling( */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 210, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 267, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 210, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 267, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_2 = ((!__pyx_t_3) != 0); - if (__pyx_t_2) { + __pyx_t_3 = ((!__pyx_t_4) != 0); + if (__pyx_t_3) { - /* "pooling_inner.pyx":211 + /* "pooling_inner.pyx":268 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< - * compute_base_sentence_pooling(&w2v, eff_sentences) - * else: + * compute_base_sentence_pooling( + * &w2v, */ { #ifdef WITH_THREAD @@ -2559,22 +2692,22 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":212 + /* "pooling_inner.pyx":269 * if not model.hierarchical: * with nogil: - * compute_base_sentence_pooling(&w2v, eff_sentences) # <<<<<<<<<<<<<< - * else: - * init_ft_s2v_config(&ft, model, target, memory) + * compute_base_sentence_pooling( # <<<<<<<<<<<<<< + * &w2v, + * eff_sentences */ __pyx_f_13pooling_inner_compute_base_sentence_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":211 + /* "pooling_inner.pyx":268 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< - * compute_base_sentence_pooling(&w2v, eff_sentences) - * else: + * compute_base_sentence_pooling( + * &w2v, */ /*finally:*/ { /*normal exit:*/{ @@ -2588,16 +2721,64 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } } - /* "pooling_inner.pyx":210 - * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + /* "pooling_inner.pyx":267 + * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: - * compute_base_sentence_pooling(&w2v, eff_sentences) + * compute_base_sentence_pooling( */ + goto __pyx_L6; } - /* "pooling_inner.pyx":205 + /* "pooling_inner.pyx":274 + * ) + * else: + * with nogil: # <<<<<<<<<<<<<< + * compute_base_sentence_hier_pooling( + * &w2v, + */ + /*else*/ { + { + #ifdef WITH_THREAD + PyThreadState *_save; + Py_UNBLOCK_THREADS + __Pyx_FastGIL_Remember(); + #endif + /*try:*/ { + + /* "pooling_inner.pyx":275 + * else: + * with nogil: + * compute_base_sentence_hier_pooling( # <<<<<<<<<<<<<< + * &w2v, + * eff_sentences, + */ + __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences, __pyx_v_window_size); + } + + /* "pooling_inner.pyx":274 + * ) + * else: + * with nogil: # <<<<<<<<<<<<<< + * compute_base_sentence_hier_pooling( + * &w2v, + */ + /*finally:*/ { + /*normal exit:*/{ + #ifdef WITH_THREAD + __Pyx_FastGIL_Forget(); + Py_BLOCK_THREADS + #endif + goto __pyx_L12; + } + __pyx_L12:; + } + } + } + __pyx_L6:; + + /* "pooling_inner.pyx":258 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< @@ -2607,101 +2788,109 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L3; } - /* "pooling_inner.pyx":214 - * compute_base_sentence_pooling(&w2v, eff_sentences) + /* "pooling_inner.pyx":281 + * ) * else: * init_ft_s2v_config(&ft, model, target, memory) # <<<<<<<<<<<<<< * - * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + * eff_sentences, eff_words = populate_ft_s2v_config( */ /*else*/ { - __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 214, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 281, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":216 - * init_ft_s2v_config(&ft, model, target, memory) - * - * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) # <<<<<<<<<<<<<< - * - * if not model.hierarchical: + /* "pooling_inner.pyx":285 + * eff_sentences, eff_words = populate_ft_s2v_config( + * &ft, + * model.wv.vocab, # <<<<<<<<<<<<<< + * indexed_sentences + * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 216, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 285, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 216, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_5); + __pyx_t_6 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 285, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 216, __pyx_L1_error) + + /* "pooling_inner.pyx":283 + * init_ft_s2v_config(&ft, model, target, memory) + * + * eff_sentences, eff_words = populate_ft_s2v_config( # <<<<<<<<<<<<<< + * &ft, + * model.wv.vocab, + */ + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_6, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 283, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; + __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { PyObject* sequence = __pyx_t_1; Py_ssize_t size = __Pyx_PySequence_SIZE(sequence); if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 216, __pyx_L1_error) + __PYX_ERR(0, 283, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { - __pyx_t_5 = PyTuple_GET_ITEM(sequence, 0); - __pyx_t_4 = PyTuple_GET_ITEM(sequence, 1); + __pyx_t_6 = PyTuple_GET_ITEM(sequence, 0); + __pyx_t_5 = PyTuple_GET_ITEM(sequence, 1); } else { - __pyx_t_5 = PyList_GET_ITEM(sequence, 0); - __pyx_t_4 = PyList_GET_ITEM(sequence, 1); + __pyx_t_6 = PyList_GET_ITEM(sequence, 0); + __pyx_t_5 = PyList_GET_ITEM(sequence, 1); } + __Pyx_INCREF(__pyx_t_6); __Pyx_INCREF(__pyx_t_5); - __Pyx_INCREF(__pyx_t_4); #else - __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 216, __pyx_L1_error) + __pyx_t_6 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 283, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_6); + __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 283, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); - __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 216, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_4); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 216, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_6); + __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 283, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_7); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; - index = 0; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L10_unpacking_failed; + __pyx_t_8 = Py_TYPE(__pyx_t_7)->tp_iternext; + index = 0; __pyx_t_6 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_6)) goto __pyx_L13_unpacking_failed; + __Pyx_GOTREF(__pyx_t_6); + index = 1; __pyx_t_5 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_5)) goto __pyx_L13_unpacking_failed; __Pyx_GOTREF(__pyx_t_5); - index = 1; __pyx_t_4 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_4)) goto __pyx_L10_unpacking_failed; - __Pyx_GOTREF(__pyx_t_4); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 216, __pyx_L1_error) - __pyx_t_7 = NULL; - __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; - goto __pyx_L11_unpacking_done; - __pyx_L10_unpacking_failed:; - __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; - __pyx_t_7 = NULL; + if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 283, __pyx_L1_error) + __pyx_t_8 = NULL; + __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; + goto __pyx_L14_unpacking_done; + __pyx_L13_unpacking_failed:; + __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; + __pyx_t_8 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 216, __pyx_L1_error) - __pyx_L11_unpacking_done:; + __PYX_ERR(0, 283, __pyx_L1_error) + __pyx_L14_unpacking_done:; } - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 216, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 283, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 283, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 216, __pyx_L1_error) - __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; __pyx_v_eff_sentences = __pyx_t_9; - __pyx_v_eff_words = __pyx_t_8; + __pyx_v_eff_words = __pyx_t_2; - /* "pooling_inner.pyx":218 - * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + /* "pooling_inner.pyx":289 + * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: * compute_ft_sentence_pooling(&ft, eff_sentences) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 218, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 289, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 218, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 289, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_3 = ((!__pyx_t_2) != 0); - if (__pyx_t_3) { + __pyx_t_4 = ((!__pyx_t_3) != 0); + if (__pyx_t_4) { - /* "pooling_inner.pyx":219 + /* "pooling_inner.pyx":290 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -2716,7 +2905,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":220 + /* "pooling_inner.pyx":291 * if not model.hierarchical: * with nogil: * compute_ft_sentence_pooling(&ft, eff_sentences) # <<<<<<<<<<<<<< @@ -2726,7 +2915,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_ft_sentence_pooling((&__pyx_v_ft), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":219 + /* "pooling_inner.pyx":290 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -2739,14 +2928,14 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_FastGIL_Forget(); Py_BLOCK_THREADS #endif - goto __pyx_L15; + goto __pyx_L18; } - __pyx_L15:; + __pyx_L18:; } } - /* "pooling_inner.pyx":218 - * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + /* "pooling_inner.pyx":289 + * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: @@ -2756,7 +2945,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } __pyx_L3:; - /* "pooling_inner.pyx":222 + /* "pooling_inner.pyx":293 * compute_ft_sentence_pooling(&ft, eff_sentences) * * return eff_sentences, eff_words # <<<<<<<<<<<<<< @@ -2764,36 +2953,36 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec * def init(): */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 222, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 293, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 222, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 222, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 293, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); + __pyx_t_6 = PyTuple_New(2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 293, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_6); __Pyx_GIVEREF(__pyx_t_1); - PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_1); - __Pyx_GIVEREF(__pyx_t_4); - PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_4); + PyTuple_SET_ITEM(__pyx_t_6, 0, __pyx_t_1); + __Pyx_GIVEREF(__pyx_t_5); + PyTuple_SET_ITEM(__pyx_t_6, 1, __pyx_t_5); __pyx_t_1 = 0; - __pyx_t_4 = 0; - __pyx_r = __pyx_t_5; __pyx_t_5 = 0; + __pyx_r = __pyx_t_6; + __pyx_t_6 = 0; goto __pyx_L0; - /* "pooling_inner.pyx":176 + /* "pooling_inner.pyx":223 * * - * def train_pooling_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< - * """Training on a sequence of sentences and update the target ndarray. - * + * def train_pooling_cy( # <<<<<<<<<<<<<< + * model, + * indexed_sentences, */ /* function exit code */ __pyx_L1_error:; __Pyx_XDECREF(__pyx_t_1); - __Pyx_XDECREF(__pyx_t_4); __Pyx_XDECREF(__pyx_t_5); __Pyx_XDECREF(__pyx_t_6); + __Pyx_XDECREF(__pyx_t_7); __Pyx_AddTraceback("pooling_inner.train_pooling_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __pyx_r = NULL; __pyx_L0:; @@ -2802,7 +2991,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec return __pyx_r; } -/* "pooling_inner.pyx":224 +/* "pooling_inner.pyx":295 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -2830,7 +3019,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("init", 0); - /* "pooling_inner.pyx":225 + /* "pooling_inner.pyx":296 * * def init(): * return 1 # <<<<<<<<<<<<<< @@ -2842,7 +3031,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __pyx_r = __pyx_int_1; goto __pyx_L0; - /* "pooling_inner.pyx":224 + /* "pooling_inner.pyx":295 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -5361,11 +5550,12 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = { {&__pyx_kp_u_unknown_dtype_code_in_numpy_pxd, __pyx_k_unknown_dtype_code_in_numpy_pxd, sizeof(__pyx_k_unknown_dtype_code_in_numpy_pxd), 0, 1, 0, 0}, {&__pyx_n_s_vocab, __pyx_k_vocab, sizeof(__pyx_k_vocab), 0, 0, 1, 1}, {&__pyx_n_s_w2v, __pyx_k_w2v, sizeof(__pyx_k_w2v), 0, 0, 1, 1}, + {&__pyx_n_s_window_size, __pyx_k_window_size, sizeof(__pyx_k_window_size), 0, 0, 1, 1}, {&__pyx_n_s_wv, __pyx_k_wv, sizeof(__pyx_k_wv), 0, 0, 1, 1}, {0, 0, 0, 0, 0, 0, 0} }; static CYTHON_SMALL_CODE int __Pyx_InitCachedBuiltins(void) { - __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 84, __pyx_L1_error) + __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 92, __pyx_L1_error) __pyx_builtin_ValueError = __Pyx_GetBuiltinName(__pyx_n_s_ValueError); if (!__pyx_builtin_ValueError) __PYX_ERR(1, 272, __pyx_L1_error) __pyx_builtin_RuntimeError = __Pyx_GetBuiltinName(__pyx_n_s_RuntimeError); if (!__pyx_builtin_RuntimeError) __PYX_ERR(1, 856, __pyx_L1_error) __pyx_builtin_ImportError = __Pyx_GetBuiltinName(__pyx_n_s_ImportError); if (!__pyx_builtin_ImportError) __PYX_ERR(1, 1038, __pyx_L1_error) @@ -5455,26 +5645,26 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { __Pyx_GOTREF(__pyx_tuple__7); __Pyx_GIVEREF(__pyx_tuple__7); - /* "pooling_inner.pyx":176 - * + /* "pooling_inner.pyx":223 * - * def train_pooling_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< - * """Training on a sequence of sentences and update the target ndarray. * + * def train_pooling_cy( # <<<<<<<<<<<<<< + * model, + * indexed_sentences, */ - __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 176, __pyx_L1_error) + __pyx_tuple__9 = PyTuple_Pack(9, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_window_size, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 223, __pyx_L1_error) __Pyx_GOTREF(__pyx_tuple__9); __Pyx_GIVEREF(__pyx_tuple__9); - __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 176, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 176, __pyx_L1_error) + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 9, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 223, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 223, __pyx_L1_error) - /* "pooling_inner.pyx":224 + /* "pooling_inner.pyx":295 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 224, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 224, __pyx_L1_error) + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 295, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 295, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -5807,88 +5997,88 @@ if (!__Pyx_RefNanny) { if (__Pyx_patch_abc() < 0) __PYX_ERR(0, 1, __pyx_L1_error) #endif - /* "pooling_inner.pyx":14 + /* "pooling_inner.pyx":15 * * import cython * import numpy as np # <<<<<<<<<<<<<< * * cimport numpy as np */ - __pyx_t_1 = __Pyx_Import(__pyx_n_s_numpy, 0, -1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 14, __pyx_L1_error) + __pyx_t_1 = __Pyx_Import(__pyx_n_s_numpy, 0, -1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 15, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_np, __pyx_t_1) < 0) __PYX_ERR(0, 14, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_np, __pyx_t_1) < 0) __PYX_ERR(0, 15, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":20 + /* "pooling_inner.pyx":21 * from libc.string cimport memset * * import scipy.linalg.blas as fblas # <<<<<<<<<<<<<< * * from average_inner cimport ( */ - __pyx_t_1 = PyList_New(1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 20, __pyx_L1_error) + __pyx_t_1 = PyList_New(1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 21, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_INCREF(__pyx_n_s__8); __Pyx_GIVEREF(__pyx_n_s__8); PyList_SET_ITEM(__pyx_t_1, 0, __pyx_n_s__8); - __pyx_t_2 = __Pyx_Import(__pyx_n_s_scipy_linalg_blas, __pyx_t_1, -1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 20, __pyx_L1_error) + __pyx_t_2 = __Pyx_Import(__pyx_n_s_scipy_linalg_blas, __pyx_t_1, -1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 21, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_2) < 0) __PYX_ERR(0, 20, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_2) < 0) __PYX_ERR(0, 21, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":176 - * + /* "pooling_inner.pyx":223 * - * def train_pooling_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< - * """Training on a sequence of sentences and update the target ndarray. * + * def train_pooling_cy( # <<<<<<<<<<<<<< + * model, + * indexed_sentences, */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 176, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 223, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 176, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 223, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":224 + /* "pooling_inner.pyx":295 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 224, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 295, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 224, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 295, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":227 + /* "pooling_inner.pyx":298 * return 1 * * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 227, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 298, __pyx_L1_error) - /* "pooling_inner.pyx":228 + /* "pooling_inner.pyx":299 * * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 228, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 299, __pyx_L1_error) - /* "pooling_inner.pyx":229 + /* "pooling_inner.pyx":300 * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() # <<<<<<<<<<<<<< */ - __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 229, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 300, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 229, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 300, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 229, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 300, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; /* "pooling_inner.pyx":1 diff --git a/fse/models/pooling_inner.pyx b/fse/models/pooling_inner.pyx index 1f51fdf..0ba50cf 100644 --- a/fse/models/pooling_inner.pyx +++ b/fse/models/pooling_inner.pyx @@ -40,7 +40,11 @@ from average_inner cimport ( DEF MAX_WORDS = 10000 DEF MAX_NGRAMS = 40 -cdef void sl_max_pool(const int *N, float *X, const float *Y) nogil: +cdef void sl_max_pool( + const int *N, + float *X, + const float *Y, +) nogil: """ Performs single left max pooling op Parameters @@ -58,7 +62,10 @@ cdef void sl_max_pool(const int *N, float *X, const float *Y) nogil: if X[i] < Y[i]: X[i] = Y[i] -cdef void compute_base_sentence_pooling(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: +cdef void compute_base_sentence_pooling( + BaseSentenceVecsConfig *c, + uINT_t num_sentences, +) nogil: """Perform optimized sentence-level max pooling for BaseAny2Vec model. Parameters @@ -83,9 +90,6 @@ cdef void compute_base_sentence_pooling(BaseSentenceVecsConfig *c, uINT_t num_se REAL_t sent_len, inv_count for sent_idx in range(num_sentences): - memset(c.mem, 0, size * cython.sizeof(REAL_t)) - memset(c.mem2, 0, size * cython.sizeof(REAL_t)) - sent_start = c.sentence_boundary[sent_idx] sent_end = c.sentence_boundary[sent_idx + 1] sent_len = ZEROF @@ -103,7 +107,49 @@ cdef void compute_base_sentence_pooling(BaseSentenceVecsConfig *c, uINT_t num_se # There's nothing to do here for many-to-one mappings -cdef void compute_ft_sentence_pooling(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: +cdef void compute_base_sentence_hier_pooling( + BaseSentenceVecsConfig *c, + uINT_t num_sentences, + uINT_t window_size, +) nogil: + """Perform optimized sentence-level hierarchical max pooling for BaseAny2Vec model. + + Parameters + ---------- + c : BaseSentenceVecsConfig * + A pointer to a fully initialized and populated struct. + num_sentences : uINT_t + The number of sentences used to train the model. + + Notes + ----- + This routine does not provide oov support. + + """ + cdef: + int size = c.size + + uINT_t sent_idx, sent_start, sent_end, sent_row + + uINT_t i, j, word_idx, word_row + + REAL_t sent_len, inv_count + + for sent_idx in range(num_sentences): + sent_start = c.sentence_boundary[sent_idx] + sent_end = c.sentence_boundary[sent_idx + 1] + sent_len = ZEROF + + for i in range(sent_start, sent_end): + sent_len += ONEF + sent_row = c.sent_adresses[i] * size + word_row = c.word_indices[i] * size + + +cdef void compute_ft_sentence_pooling( + FTSentenceVecsConfig *c, + uINT_t num_sentences, +) nogil: """Perform optimized sentence-level max pooling for FastText model. Parameters @@ -174,7 +220,12 @@ cdef void compute_ft_sentence_pooling(FTSentenceVecsConfig *c, uINT_t num_senten # There's nothing to do here for many-to-one mappings -def train_pooling_cy(model, indexed_sentences, target, memory): +def train_pooling_cy( + model, + indexed_sentences, + target, + memory +): """Training on a sequence of sentences and update the target ndarray. Called internally from :meth:`~fse.models.pooling.MaxPooling._do_train_job`. @@ -200,21 +251,40 @@ def train_pooling_cy(model, indexed_sentences, target, memory): cdef uINT_t eff_sentences = 0 cdef uINT_t eff_words = 0 + cdef uINT_t window_size = model.window_size cdef BaseSentenceVecsConfig w2v cdef FTSentenceVecsConfig ft if not model.is_ft: init_base_s2v_config(&w2v, model, target, memory) - eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + eff_sentences, eff_words = populate_base_s2v_config( + &w2v, + model.wv.vocab, + indexed_sentences + ) if not model.hierarchical: - with nogil: - compute_base_sentence_pooling(&w2v, eff_sentences) + with nogil: + compute_base_sentence_pooling( + &w2v, + eff_sentences + ) + else: + with nogil: + compute_base_sentence_hier_pooling( + &w2v, + eff_sentences, + window_size + ) else: init_ft_s2v_config(&ft, model, target, memory) - eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + eff_sentences, eff_words = populate_ft_s2v_config( + &ft, + model.wv.vocab, + indexed_sentences + ) if not model.hierarchical: with nogil: From 3305b72715b9d250e1b59b2c35dda435366c7230 Mon Sep 17 00:00:00 2001 From: OliverB Date: Thu, 27 Feb 2020 18:55:40 +0100 Subject: [PATCH 29/63] Refactoring --- fse/models/average_inner.c | 960 ++++++++++++++++++----------------- fse/models/average_inner.pxd | 45 +- fse/models/average_inner.pyx | 74 ++- fse/models/pooling_inner.pyx | 1 - 4 files changed, 591 insertions(+), 489 deletions(-) diff --git a/fse/models/average_inner.c b/fse/models/average_inner.c index 91211b6..cb3ff81 100644 --- a/fse/models/average_inner.c +++ b/fse/models/average_inner.c @@ -1053,7 +1053,7 @@ typedef npy_double __pyx_t_5numpy_double_t; */ typedef npy_longdouble __pyx_t_5numpy_longdouble_t; -/* "average_inner.pxd":15 +/* "average_inner.pxd":16 * void* PyCObject_AsVoidPtr(object obj) * * ctypedef np.float32_t REAL_t # <<<<<<<<<<<<<< @@ -1062,7 +1062,7 @@ typedef npy_longdouble __pyx_t_5numpy_longdouble_t; */ typedef __pyx_t_5numpy_float32_t __pyx_t_13average_inner_REAL_t; -/* "average_inner.pxd":16 +/* "average_inner.pxd":17 * * ctypedef np.float32_t REAL_t * ctypedef np.uint32_t uINT_t # <<<<<<<<<<<<<< @@ -1135,25 +1135,25 @@ typedef npy_cdouble __pyx_t_5numpy_complex_t; struct __pyx_t_13average_inner_BaseSentenceVecsConfig; struct __pyx_t_13average_inner_FTSentenceVecsConfig; -/* "average_inner.pxd":19 +/* "average_inner.pxd":20 * * # BLAS routine signatures - * ctypedef void (*saxpy_ptr) (const int *N, const float *alpha, const float *X, const int *incX, float *Y, const int *incY) nogil # <<<<<<<<<<<<<< - * ctypedef void (*sscal_ptr) (const int *N, const float *alpha, const float *X, const int *incX) nogil - * + * ctypedef void (*saxpy_ptr) ( # <<<<<<<<<<<<<< + * const int *N, + * const float *alpha, */ typedef void (*__pyx_t_13average_inner_saxpy_ptr)(int const *, float const *, float const *, int const *, float *, int const *); -/* "average_inner.pxd":20 - * # BLAS routine signatures - * ctypedef void (*saxpy_ptr) (const int *N, const float *alpha, const float *X, const int *incX, float *Y, const int *incY) nogil - * ctypedef void (*sscal_ptr) (const int *N, const float *alpha, const float *X, const int *incX) nogil # <<<<<<<<<<<<<< +/* "average_inner.pxd":29 + * ) nogil * - * cdef saxpy_ptr saxpy + * ctypedef void (*sscal_ptr) ( # <<<<<<<<<<<<<< + * const int *N, + * const float *alpha, */ typedef void (*__pyx_t_13average_inner_sscal_ptr)(int const *, float const *, float const *, int const *); -/* "average_inner.pxd":34 +/* "average_inner.pxd":48 * DEF MAX_NGRAMS = 40 * * cdef struct BaseSentenceVecsConfig: # <<<<<<<<<<<<<< @@ -1173,7 +1173,7 @@ struct __pyx_t_13average_inner_BaseSentenceVecsConfig { __pyx_t_13average_inner_uINT_t sentence_boundary[(0x2710 + 1)]; }; -/* "average_inner.pxd":48 +/* "average_inner.pxd":62 * uINT_t sentence_boundary[MAX_WORDS + 1] * * cdef struct FTSentenceVecsConfig: # <<<<<<<<<<<<<< @@ -1959,12 +1959,12 @@ static PyObject *__pyx_codeobj__10; static PyObject *__pyx_codeobj__11; /* Late includes */ -/* "average_inner.pyx":36 +/* "average_inner.pyx":37 * DEF MAX_NGRAMS = 40 * - * cdef init_base_s2v_config(BaseSentenceVecsConfig *c, model, target, memory): # <<<<<<<<<<<<<< - * """Load BaseAny2Vec parameters into a BaseSentenceVecsConfig struct. - * + * cdef init_base_s2v_config( # <<<<<<<<<<<<<< + * BaseSentenceVecsConfig *c, + * model, */ static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, PyObject *__pyx_v_model, PyObject *__pyx_v_target, PyObject *__pyx_v_memory) { @@ -1975,106 +1975,106 @@ static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13a PyObject *__pyx_t_3 = NULL; __Pyx_RefNannySetupContext("init_base_s2v_config", 0); - /* "average_inner.pyx":52 + /* "average_inner.pyx":58 * * """ * c[0].workers = model.workers # <<<<<<<<<<<<<< * c[0].size = model.sv.vector_size * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_workers); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 52, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_workers); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 58, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 52, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 58, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; (__pyx_v_c[0]).workers = __pyx_t_2; - /* "average_inner.pyx":53 + /* "average_inner.pyx":59 * """ * c[0].workers = model.workers * c[0].size = model.sv.vector_size # <<<<<<<<<<<<<< * * c[0].mem = (np.PyArray_DATA(memory[0])) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_sv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 53, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_sv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 59, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vector_size); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 53, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vector_size); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 59, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 53, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 59, __pyx_L1_error) __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; (__pyx_v_c[0]).size = __pyx_t_2; - /* "average_inner.pyx":55 + /* "average_inner.pyx":61 * c[0].size = model.sv.vector_size * * c[0].mem = (np.PyArray_DATA(memory[0])) # <<<<<<<<<<<<<< * c[0].mem2 = (np.PyArray_DATA(memory[2])) * */ - __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_memory, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 55, __pyx_L1_error) + __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_memory, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 61, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 55, __pyx_L1_error) + if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 61, __pyx_L1_error) (__pyx_v_c[0]).mem = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_3))); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - /* "average_inner.pyx":56 + /* "average_inner.pyx":62 * * c[0].mem = (np.PyArray_DATA(memory[0])) * c[0].mem2 = (np.PyArray_DATA(memory[2])) # <<<<<<<<<<<<<< * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors)) */ - __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 56, __pyx_L1_error) + __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 62, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 56, __pyx_L1_error) + if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 62, __pyx_L1_error) (__pyx_v_c[0]).mem2 = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_3))); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - /* "average_inner.pyx":58 + /* "average_inner.pyx":64 * c[0].mem2 = (np.PyArray_DATA(memory[2])) * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors)) # <<<<<<<<<<<<<< * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) * */ - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 58, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 64, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_vectors); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 58, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_vectors); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 64, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 58, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 64, __pyx_L1_error) (__pyx_v_c[0]).word_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":59 + /* "average_inner.pyx":65 * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors)) * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) # <<<<<<<<<<<<<< * * c[0].sentence_vectors = (np.PyArray_DATA(target)) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 59, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 65, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 59, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 65, __pyx_L1_error) (__pyx_v_c[0]).word_weights = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":61 + /* "average_inner.pyx":67 * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) * * c[0].sentence_vectors = (np.PyArray_DATA(target)) # <<<<<<<<<<<<<< * - * cdef init_ft_s2v_config(FTSentenceVecsConfig *c, model, target, memory): + * cdef init_ft_s2v_config( */ - if (!(likely(((__pyx_v_target) == Py_None) || likely(__Pyx_TypeTest(__pyx_v_target, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 61, __pyx_L1_error) + if (!(likely(((__pyx_v_target) == Py_None) || likely(__Pyx_TypeTest(__pyx_v_target, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 67, __pyx_L1_error) (__pyx_v_c[0]).sentence_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_v_target))); - /* "average_inner.pyx":36 + /* "average_inner.pyx":37 * DEF MAX_NGRAMS = 40 * - * cdef init_base_s2v_config(BaseSentenceVecsConfig *c, model, target, memory): # <<<<<<<<<<<<<< - * """Load BaseAny2Vec parameters into a BaseSentenceVecsConfig struct. - * + * cdef init_base_s2v_config( # <<<<<<<<<<<<<< + * BaseSentenceVecsConfig *c, + * model, */ /* function exit code */ @@ -2091,12 +2091,12 @@ static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13a return __pyx_r; } -/* "average_inner.pyx":63 +/* "average_inner.pyx":69 * c[0].sentence_vectors = (np.PyArray_DATA(target)) * - * cdef init_ft_s2v_config(FTSentenceVecsConfig *c, model, target, memory): # <<<<<<<<<<<<<< - * """Load Fasttext parameters into a FTSentenceVecsConfig struct. - * + * cdef init_ft_s2v_config( # <<<<<<<<<<<<<< + * FTSentenceVecsConfig *c, + * model, */ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, PyObject *__pyx_v_model, PyObject *__pyx_v_target, PyObject *__pyx_v_memory) { @@ -2110,96 +2110,96 @@ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13ave __pyx_t_13average_inner_REAL_t __pyx_t_6; __Pyx_RefNannySetupContext("init_ft_s2v_config", 0); - /* "average_inner.pyx":80 + /* "average_inner.pyx":91 * """ * * c[0].workers = model.workers # <<<<<<<<<<<<<< * c[0].size = model.sv.vector_size * c[0].min_n = model.wv.min_n */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_workers); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 80, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_workers); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 91, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 80, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 91, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; (__pyx_v_c[0]).workers = __pyx_t_2; - /* "average_inner.pyx":81 + /* "average_inner.pyx":92 * * c[0].workers = model.workers * c[0].size = model.sv.vector_size # <<<<<<<<<<<<<< * c[0].min_n = model.wv.min_n * c[0].max_n = model.wv.max_n */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_sv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 81, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_sv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 92, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vector_size); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 81, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vector_size); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 92, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 81, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 92, __pyx_L1_error) __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; (__pyx_v_c[0]).size = __pyx_t_2; - /* "average_inner.pyx":82 + /* "average_inner.pyx":93 * c[0].workers = model.workers * c[0].size = model.sv.vector_size * c[0].min_n = model.wv.min_n # <<<<<<<<<<<<<< * c[0].max_n = model.wv.max_n * c[0].bucket = model.wv.bucket */ - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 82, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 93, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_min_n); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 82, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_min_n); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 93, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 82, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 93, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; (__pyx_v_c[0]).min_n = __pyx_t_2; - /* "average_inner.pyx":83 + /* "average_inner.pyx":94 * c[0].size = model.sv.vector_size * c[0].min_n = model.wv.min_n * c[0].max_n = model.wv.max_n # <<<<<<<<<<<<<< * c[0].bucket = model.wv.bucket * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 83, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 94, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_max_n); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 83, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_max_n); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 94, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 83, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 94, __pyx_L1_error) __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; (__pyx_v_c[0]).max_n = __pyx_t_2; - /* "average_inner.pyx":84 + /* "average_inner.pyx":95 * c[0].min_n = model.wv.min_n * c[0].max_n = model.wv.max_n * c[0].bucket = model.wv.bucket # <<<<<<<<<<<<<< * * c[0].oov_weight = np.max(model.word_weights) */ - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 84, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 95, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_bucket); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 84, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_bucket); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 95, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 84, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 95, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; (__pyx_v_c[0]).bucket = __pyx_t_2; - /* "average_inner.pyx":86 + /* "average_inner.pyx":97 * c[0].bucket = model.wv.bucket * * c[0].oov_weight = np.max(model.word_weights) # <<<<<<<<<<<<<< * * c[0].mem = (np.PyArray_DATA(memory[0])) */ - __Pyx_GetModuleGlobalName(__pyx_t_3, __pyx_n_s_np); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 86, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_3, __pyx_n_s_np); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 97, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_max); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 86, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_max); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 97, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 86, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 97, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); __pyx_t_5 = NULL; if (CYTHON_UNPACK_METHODS && unlikely(PyMethod_Check(__pyx_t_4))) { @@ -2214,52 +2214,52 @@ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13ave __pyx_t_1 = (__pyx_t_5) ? __Pyx_PyObject_Call2Args(__pyx_t_4, __pyx_t_5, __pyx_t_3) : __Pyx_PyObject_CallOneArg(__pyx_t_4, __pyx_t_3); __Pyx_XDECREF(__pyx_t_5); __pyx_t_5 = 0; __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 86, __pyx_L1_error) + if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 97, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; - __pyx_t_6 = __pyx_PyFloat_AsFloat(__pyx_t_1); if (unlikely((__pyx_t_6 == ((npy_float32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 86, __pyx_L1_error) + __pyx_t_6 = __pyx_PyFloat_AsFloat(__pyx_t_1); if (unlikely((__pyx_t_6 == ((npy_float32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 97, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; (__pyx_v_c[0]).oov_weight = ((__pyx_t_13average_inner_REAL_t)__pyx_t_6); - /* "average_inner.pyx":88 + /* "average_inner.pyx":99 * c[0].oov_weight = np.max(model.word_weights) * * c[0].mem = (np.PyArray_DATA(memory[0])) # <<<<<<<<<<<<<< * c[0].mem2 = (np.PyArray_DATA(memory[2])) * */ - __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 88, __pyx_L1_error) + __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 99, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 88, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 99, __pyx_L1_error) (__pyx_v_c[0]).mem = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":89 + /* "average_inner.pyx":100 * * c[0].mem = (np.PyArray_DATA(memory[0])) * c[0].mem2 = (np.PyArray_DATA(memory[2])) # <<<<<<<<<<<<<< * * memory[1].fill(ZERO) # Reset the ngram storage before filling the struct */ - __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 89, __pyx_L1_error) + __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 100, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 89, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 100, __pyx_L1_error) (__pyx_v_c[0]).mem2 = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":91 + /* "average_inner.pyx":102 * c[0].mem2 = (np.PyArray_DATA(memory[2])) * * memory[1].fill(ZERO) # Reset the ngram storage before filling the struct # <<<<<<<<<<<<<< * c[0].subwords_idx = (np.PyArray_DATA(memory[1])) * */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 91, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 102, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_4, __pyx_n_s_fill); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 91, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_4, __pyx_n_s_fill); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 102, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; - __pyx_t_4 = __Pyx_PyInt_From_int(__pyx_v_13average_inner_ZERO); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 91, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyInt_From_int(__pyx_v_13average_inner_ZERO); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 102, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __pyx_t_5 = NULL; if (CYTHON_UNPACK_METHODS && likely(PyMethod_Check(__pyx_t_3))) { @@ -2274,85 +2274,85 @@ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13ave __pyx_t_1 = (__pyx_t_5) ? __Pyx_PyObject_Call2Args(__pyx_t_3, __pyx_t_5, __pyx_t_4) : __Pyx_PyObject_CallOneArg(__pyx_t_3, __pyx_t_4); __Pyx_XDECREF(__pyx_t_5); __pyx_t_5 = 0; __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; - if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 91, __pyx_L1_error) + if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 102, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":92 + /* "average_inner.pyx":103 * * memory[1].fill(ZERO) # Reset the ngram storage before filling the struct * c[0].subwords_idx = (np.PyArray_DATA(memory[1])) # <<<<<<<<<<<<<< * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) */ - __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 92, __pyx_L1_error) + __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 103, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 92, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 103, __pyx_L1_error) (__pyx_v_c[0]).subwords_idx = ((__pyx_t_13average_inner_uINT_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":94 + /* "average_inner.pyx":105 * c[0].subwords_idx = (np.PyArray_DATA(memory[1])) * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) # <<<<<<<<<<<<<< * c[0].ngram_vectors = (np.PyArray_DATA(model.wv.vectors_ngrams)) * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 94, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 105, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vectors_vocab); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 94, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vectors_vocab); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 105, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 94, __pyx_L1_error) + if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 105, __pyx_L1_error) (__pyx_v_c[0]).word_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_3))); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - /* "average_inner.pyx":95 + /* "average_inner.pyx":106 * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) * c[0].ngram_vectors = (np.PyArray_DATA(model.wv.vectors_ngrams)) # <<<<<<<<<<<<<< * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) * */ - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 95, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 106, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_vectors_ngrams); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 95, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_vectors_ngrams); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 106, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 95, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 106, __pyx_L1_error) (__pyx_v_c[0]).ngram_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":96 + /* "average_inner.pyx":107 * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) * c[0].ngram_vectors = (np.PyArray_DATA(model.wv.vectors_ngrams)) * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) # <<<<<<<<<<<<<< * * c[0].sentence_vectors = (np.PyArray_DATA(target)) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 96, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 107, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 96, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 107, __pyx_L1_error) (__pyx_v_c[0]).word_weights = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":98 + /* "average_inner.pyx":109 * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) * * c[0].sentence_vectors = (np.PyArray_DATA(target)) # <<<<<<<<<<<<<< * - * cdef object populate_base_s2v_config(BaseSentenceVecsConfig *c, vocab, indexed_sentences): + * cdef object populate_base_s2v_config( */ - if (!(likely(((__pyx_v_target) == Py_None) || likely(__Pyx_TypeTest(__pyx_v_target, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 98, __pyx_L1_error) + if (!(likely(((__pyx_v_target) == Py_None) || likely(__Pyx_TypeTest(__pyx_v_target, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 109, __pyx_L1_error) (__pyx_v_c[0]).sentence_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_v_target))); - /* "average_inner.pyx":63 + /* "average_inner.pyx":69 * c[0].sentence_vectors = (np.PyArray_DATA(target)) * - * cdef init_ft_s2v_config(FTSentenceVecsConfig *c, model, target, memory): # <<<<<<<<<<<<<< - * """Load Fasttext parameters into a FTSentenceVecsConfig struct. - * + * cdef init_ft_s2v_config( # <<<<<<<<<<<<<< + * FTSentenceVecsConfig *c, + * model, */ /* function exit code */ @@ -2371,12 +2371,12 @@ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13ave return __pyx_r; } -/* "average_inner.pyx":100 +/* "average_inner.pyx":111 * c[0].sentence_vectors = (np.PyArray_DATA(target)) * - * cdef object populate_base_s2v_config(BaseSentenceVecsConfig *c, vocab, indexed_sentences): # <<<<<<<<<<<<<< - * """Prepare C structures for BaseAny2VecModel so we can go "full C" and release the Python GIL. - * + * cdef object populate_base_s2v_config( # <<<<<<<<<<<<<< + * BaseSentenceVecsConfig *c, + * vocab, */ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, PyObject *__pyx_v_vocab, PyObject *__pyx_v_indexed_sentences) { @@ -2400,7 +2400,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_t_13average_inner_uINT_t __pyx_t_11; __Pyx_RefNannySetupContext("populate_base_s2v_config", 0); - /* "average_inner.pyx":124 + /* "average_inner.pyx":139 * """ * * cdef uINT_t eff_words = ZERO # Effective words encountered in a sentence # <<<<<<<<<<<<<< @@ -2409,7 +2409,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ __pyx_v_eff_words = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":125 + /* "average_inner.pyx":140 * * cdef uINT_t eff_words = ZERO # Effective words encountered in a sentence * cdef uINT_t eff_sents = ZERO # Effective sentences encountered # <<<<<<<<<<<<<< @@ -2418,7 +2418,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ __pyx_v_eff_sents = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":127 + /* "average_inner.pyx":142 * cdef uINT_t eff_sents = ZERO # Effective sentences encountered * * c.sentence_boundary[0] = ZERO # <<<<<<<<<<<<<< @@ -2427,7 +2427,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ (__pyx_v_c->sentence_boundary[0]) = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":129 + /* "average_inner.pyx":144 * c.sentence_boundary[0] = ZERO * * for obj in indexed_sentences: # <<<<<<<<<<<<<< @@ -2438,26 +2438,26 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_t_1 = __pyx_v_indexed_sentences; __Pyx_INCREF(__pyx_t_1); __pyx_t_2 = 0; __pyx_t_3 = NULL; } else { - __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 129, __pyx_L1_error) + __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 144, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext; if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 129, __pyx_L1_error) + __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext; if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 144, __pyx_L1_error) } for (;;) { if (likely(!__pyx_t_3)) { if (likely(PyList_CheckExact(__pyx_t_1))) { if (__pyx_t_2 >= PyList_GET_SIZE(__pyx_t_1)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 129, __pyx_L1_error) + __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 144, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 129, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 144, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } else { if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_1)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 129, __pyx_L1_error) + __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 144, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 129, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 144, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } @@ -2467,7 +2467,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t PyObject* exc_type = PyErr_Occurred(); if (exc_type) { if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); - else __PYX_ERR(0, 129, __pyx_L1_error) + else __PYX_ERR(0, 144, __pyx_L1_error) } break; } @@ -2476,21 +2476,21 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __Pyx_XDECREF_SET(__pyx_v_obj, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":130 + /* "average_inner.pyx":145 * * for obj in indexed_sentences: * if not obj[0]: # <<<<<<<<<<<<<< * continue * for token in obj[0]: */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 130, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 145, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 130, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 145, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; __pyx_t_6 = ((!__pyx_t_5) != 0); if (__pyx_t_6) { - /* "average_inner.pyx":131 + /* "average_inner.pyx":146 * for obj in indexed_sentences: * if not obj[0]: * continue # <<<<<<<<<<<<<< @@ -2499,7 +2499,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ goto __pyx_L3_continue; - /* "average_inner.pyx":130 + /* "average_inner.pyx":145 * * for obj in indexed_sentences: * if not obj[0]: # <<<<<<<<<<<<<< @@ -2508,22 +2508,22 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ } - /* "average_inner.pyx":132 + /* "average_inner.pyx":147 * if not obj[0]: * continue * for token in obj[0]: # <<<<<<<<<<<<<< * word = vocab[token] if token in vocab else None # Vocab obj * if word is None: */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 132, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 147, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); if (likely(PyList_CheckExact(__pyx_t_4)) || PyTuple_CheckExact(__pyx_t_4)) { __pyx_t_7 = __pyx_t_4; __Pyx_INCREF(__pyx_t_7); __pyx_t_8 = 0; __pyx_t_9 = NULL; } else { - __pyx_t_8 = -1; __pyx_t_7 = PyObject_GetIter(__pyx_t_4); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 132, __pyx_L1_error) + __pyx_t_8 = -1; __pyx_t_7 = PyObject_GetIter(__pyx_t_4); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 147, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); - __pyx_t_9 = Py_TYPE(__pyx_t_7)->tp_iternext; if (unlikely(!__pyx_t_9)) __PYX_ERR(0, 132, __pyx_L1_error) + __pyx_t_9 = Py_TYPE(__pyx_t_7)->tp_iternext; if (unlikely(!__pyx_t_9)) __PYX_ERR(0, 147, __pyx_L1_error) } __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; for (;;) { @@ -2531,17 +2531,17 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t if (likely(PyList_CheckExact(__pyx_t_7))) { if (__pyx_t_8 >= PyList_GET_SIZE(__pyx_t_7)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyList_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 132, __pyx_L1_error) + __pyx_t_4 = PyList_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 147, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 132, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 147, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } else { if (__pyx_t_8 >= PyTuple_GET_SIZE(__pyx_t_7)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 132, __pyx_L1_error) + __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 147, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 132, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 147, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } @@ -2551,7 +2551,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t PyObject* exc_type = PyErr_Occurred(); if (exc_type) { if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); - else __PYX_ERR(0, 132, __pyx_L1_error) + else __PYX_ERR(0, 147, __pyx_L1_error) } break; } @@ -2560,16 +2560,16 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __Pyx_XDECREF_SET(__pyx_v_token, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":133 + /* "average_inner.pyx":148 * continue * for token in obj[0]: * word = vocab[token] if token in vocab else None # Vocab obj # <<<<<<<<<<<<<< * if word is None: * continue */ - __pyx_t_6 = (__Pyx_PySequence_ContainsTF(__pyx_v_token, __pyx_v_vocab, Py_EQ)); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(0, 133, __pyx_L1_error) + __pyx_t_6 = (__Pyx_PySequence_ContainsTF(__pyx_v_token, __pyx_v_vocab, Py_EQ)); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(0, 148, __pyx_L1_error) if ((__pyx_t_6 != 0)) { - __pyx_t_10 = __Pyx_PyObject_GetItem(__pyx_v_vocab, __pyx_v_token); if (unlikely(!__pyx_t_10)) __PYX_ERR(0, 133, __pyx_L1_error) + __pyx_t_10 = __Pyx_PyObject_GetItem(__pyx_v_vocab, __pyx_v_token); if (unlikely(!__pyx_t_10)) __PYX_ERR(0, 148, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_10); __pyx_t_4 = __pyx_t_10; __pyx_t_10 = 0; @@ -2580,7 +2580,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __Pyx_XDECREF_SET(__pyx_v_word, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":134 + /* "average_inner.pyx":149 * for token in obj[0]: * word = vocab[token] if token in vocab else None # Vocab obj * if word is None: # <<<<<<<<<<<<<< @@ -2591,7 +2591,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_t_5 = (__pyx_t_6 != 0); if (__pyx_t_5) { - /* "average_inner.pyx":135 + /* "average_inner.pyx":150 * word = vocab[token] if token in vocab else None # Vocab obj * if word is None: * continue # <<<<<<<<<<<<<< @@ -2600,7 +2600,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ goto __pyx_L6_continue; - /* "average_inner.pyx":134 + /* "average_inner.pyx":149 * for token in obj[0]: * word = vocab[token] if token in vocab else None # Vocab obj * if word is None: # <<<<<<<<<<<<<< @@ -2609,33 +2609,33 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ } - /* "average_inner.pyx":136 + /* "average_inner.pyx":151 * if word is None: * continue * c.word_indices[eff_words] = word.index # <<<<<<<<<<<<<< * c.sent_adresses[eff_words] = obj[1] * */ - __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_v_word, __pyx_n_s_index); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 136, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_v_word, __pyx_n_s_index); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 151, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_11 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_11 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 136, __pyx_L1_error) + __pyx_t_11 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_11 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 151, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; (__pyx_v_c->word_indices[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_11); - /* "average_inner.pyx":137 + /* "average_inner.pyx":152 * continue * c.word_indices[eff_words] = word.index * c.sent_adresses[eff_words] = obj[1] # <<<<<<<<<<<<<< * * eff_words += ONE */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 137, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 152, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_11 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_11 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 137, __pyx_L1_error) + __pyx_t_11 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_11 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 152, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; (__pyx_v_c->sent_adresses[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_11); - /* "average_inner.pyx":139 + /* "average_inner.pyx":154 * c.sent_adresses[eff_words] = obj[1] * * eff_words += ONE # <<<<<<<<<<<<<< @@ -2644,7 +2644,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ __pyx_v_eff_words = (__pyx_v_eff_words + __pyx_v_13average_inner_ONE); - /* "average_inner.pyx":140 + /* "average_inner.pyx":155 * * eff_words += ONE * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -2654,7 +2654,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_t_5 = ((__pyx_v_eff_words == 0x2710) != 0); if (__pyx_t_5) { - /* "average_inner.pyx":141 + /* "average_inner.pyx":156 * eff_words += ONE * if eff_words == MAX_WORDS: * break # <<<<<<<<<<<<<< @@ -2663,7 +2663,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ goto __pyx_L7_break; - /* "average_inner.pyx":140 + /* "average_inner.pyx":155 * * eff_words += ONE * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -2672,7 +2672,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ } - /* "average_inner.pyx":132 + /* "average_inner.pyx":147 * if not obj[0]: * continue * for token in obj[0]: # <<<<<<<<<<<<<< @@ -2684,7 +2684,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_L7_break:; __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; - /* "average_inner.pyx":142 + /* "average_inner.pyx":157 * if eff_words == MAX_WORDS: * break * eff_sents += 1 # <<<<<<<<<<<<<< @@ -2693,7 +2693,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ __pyx_v_eff_sents = (__pyx_v_eff_sents + 1); - /* "average_inner.pyx":143 + /* "average_inner.pyx":158 * break * eff_sents += 1 * c.sentence_boundary[eff_sents] = eff_words # <<<<<<<<<<<<<< @@ -2702,7 +2702,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ (__pyx_v_c->sentence_boundary[__pyx_v_eff_sents]) = __pyx_v_eff_words; - /* "average_inner.pyx":145 + /* "average_inner.pyx":160 * c.sentence_boundary[eff_sents] = eff_words * * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -2712,7 +2712,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_t_5 = ((__pyx_v_eff_words == 0x2710) != 0); if (__pyx_t_5) { - /* "average_inner.pyx":146 + /* "average_inner.pyx":161 * * if eff_words == MAX_WORDS: * break # <<<<<<<<<<<<<< @@ -2721,7 +2721,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ goto __pyx_L4_break; - /* "average_inner.pyx":145 + /* "average_inner.pyx":160 * c.sentence_boundary[eff_sents] = eff_words * * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -2730,7 +2730,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ } - /* "average_inner.pyx":129 + /* "average_inner.pyx":144 * c.sentence_boundary[0] = ZERO * * for obj in indexed_sentences: # <<<<<<<<<<<<<< @@ -2742,19 +2742,19 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_L4_break:; __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":148 + /* "average_inner.pyx":163 * break * * return eff_sents, eff_words # <<<<<<<<<<<<<< * - * cdef object populate_ft_s2v_config(FTSentenceVecsConfig *c, vocab, indexed_sentences): + * cdef object populate_ft_s2v_config( */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sents); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 148, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sents); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 163, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_7 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 148, __pyx_L1_error) + __pyx_t_7 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 163, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); - __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 148, __pyx_L1_error) + __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 163, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_1); @@ -2766,12 +2766,12 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_t_4 = 0; goto __pyx_L0; - /* "average_inner.pyx":100 + /* "average_inner.pyx":111 * c[0].sentence_vectors = (np.PyArray_DATA(target)) * - * cdef object populate_base_s2v_config(BaseSentenceVecsConfig *c, vocab, indexed_sentences): # <<<<<<<<<<<<<< - * """Prepare C structures for BaseAny2VecModel so we can go "full C" and release the Python GIL. - * + * cdef object populate_base_s2v_config( # <<<<<<<<<<<<<< + * BaseSentenceVecsConfig *c, + * vocab, */ /* function exit code */ @@ -2791,12 +2791,12 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t return __pyx_r; } -/* "average_inner.pyx":150 +/* "average_inner.pyx":165 * return eff_sents, eff_words * - * cdef object populate_ft_s2v_config(FTSentenceVecsConfig *c, vocab, indexed_sentences): # <<<<<<<<<<<<<< - * """Prepare C structures for FastText so we can go "full C" and release the Python GIL. - * + * cdef object populate_ft_s2v_config( # <<<<<<<<<<<<<< + * FTSentenceVecsConfig *c, + * vocab, */ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, PyObject *__pyx_v_vocab, PyObject *__pyx_v_indexed_sentences) { @@ -2834,7 +2834,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 Py_ssize_t __pyx_t_20; __Pyx_RefNannySetupContext("populate_ft_s2v_config", 0); - /* "average_inner.pyx":174 + /* "average_inner.pyx":193 * """ * * cdef uINT_t eff_words = ZERO # Effective words encountered in a sentence # <<<<<<<<<<<<<< @@ -2843,7 +2843,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ __pyx_v_eff_words = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":175 + /* "average_inner.pyx":194 * * cdef uINT_t eff_words = ZERO # Effective words encountered in a sentence * cdef uINT_t eff_sents = ZERO # Effective sentences encountered # <<<<<<<<<<<<<< @@ -2852,7 +2852,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ __pyx_v_eff_sents = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":177 + /* "average_inner.pyx":196 * cdef uINT_t eff_sents = ZERO # Effective sentences encountered * * c.sentence_boundary[0] = ZERO # <<<<<<<<<<<<<< @@ -2861,7 +2861,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ (__pyx_v_c->sentence_boundary[0]) = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":179 + /* "average_inner.pyx":198 * c.sentence_boundary[0] = ZERO * * for obj in indexed_sentences: # <<<<<<<<<<<<<< @@ -2872,26 +2872,26 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __pyx_t_1 = __pyx_v_indexed_sentences; __Pyx_INCREF(__pyx_t_1); __pyx_t_2 = 0; __pyx_t_3 = NULL; } else { - __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 179, __pyx_L1_error) + __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 198, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext; if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 179, __pyx_L1_error) + __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext; if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 198, __pyx_L1_error) } for (;;) { if (likely(!__pyx_t_3)) { if (likely(PyList_CheckExact(__pyx_t_1))) { if (__pyx_t_2 >= PyList_GET_SIZE(__pyx_t_1)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 179, __pyx_L1_error) + __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 198, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 179, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 198, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } else { if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_1)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 179, __pyx_L1_error) + __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 198, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 179, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 198, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } @@ -2901,7 +2901,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 PyObject* exc_type = PyErr_Occurred(); if (exc_type) { if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); - else __PYX_ERR(0, 179, __pyx_L1_error) + else __PYX_ERR(0, 198, __pyx_L1_error) } break; } @@ -2910,21 +2910,21 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __Pyx_XDECREF_SET(__pyx_v_obj, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":180 + /* "average_inner.pyx":199 * * for obj in indexed_sentences: * if not obj[0]: # <<<<<<<<<<<<<< * continue * for token in obj[0]: */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 180, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 199, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 180, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 199, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; __pyx_t_6 = ((!__pyx_t_5) != 0); if (__pyx_t_6) { - /* "average_inner.pyx":181 + /* "average_inner.pyx":200 * for obj in indexed_sentences: * if not obj[0]: * continue # <<<<<<<<<<<<<< @@ -2933,7 +2933,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ goto __pyx_L3_continue; - /* "average_inner.pyx":180 + /* "average_inner.pyx":199 * * for obj in indexed_sentences: * if not obj[0]: # <<<<<<<<<<<<<< @@ -2942,22 +2942,22 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ } - /* "average_inner.pyx":182 + /* "average_inner.pyx":201 * if not obj[0]: * continue * for token in obj[0]: # <<<<<<<<<<<<<< * c.sent_adresses[eff_words] = obj[1] * if token in vocab: */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 182, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 201, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); if (likely(PyList_CheckExact(__pyx_t_4)) || PyTuple_CheckExact(__pyx_t_4)) { __pyx_t_7 = __pyx_t_4; __Pyx_INCREF(__pyx_t_7); __pyx_t_8 = 0; __pyx_t_9 = NULL; } else { - __pyx_t_8 = -1; __pyx_t_7 = PyObject_GetIter(__pyx_t_4); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 182, __pyx_L1_error) + __pyx_t_8 = -1; __pyx_t_7 = PyObject_GetIter(__pyx_t_4); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 201, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); - __pyx_t_9 = Py_TYPE(__pyx_t_7)->tp_iternext; if (unlikely(!__pyx_t_9)) __PYX_ERR(0, 182, __pyx_L1_error) + __pyx_t_9 = Py_TYPE(__pyx_t_7)->tp_iternext; if (unlikely(!__pyx_t_9)) __PYX_ERR(0, 201, __pyx_L1_error) } __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; for (;;) { @@ -2965,17 +2965,17 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 if (likely(PyList_CheckExact(__pyx_t_7))) { if (__pyx_t_8 >= PyList_GET_SIZE(__pyx_t_7)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyList_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 182, __pyx_L1_error) + __pyx_t_4 = PyList_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 201, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 182, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 201, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } else { if (__pyx_t_8 >= PyTuple_GET_SIZE(__pyx_t_7)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 182, __pyx_L1_error) + __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 201, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 182, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 201, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } @@ -2985,7 +2985,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 PyObject* exc_type = PyErr_Occurred(); if (exc_type) { if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); - else __PYX_ERR(0, 182, __pyx_L1_error) + else __PYX_ERR(0, 201, __pyx_L1_error) } break; } @@ -2994,56 +2994,56 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __Pyx_XDECREF_SET(__pyx_v_token, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":183 + /* "average_inner.pyx":202 * continue * for token in obj[0]: * c.sent_adresses[eff_words] = obj[1] # <<<<<<<<<<<<<< * if token in vocab: * # In Vocabulary */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 183, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 202, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 183, __pyx_L1_error) + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 202, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; (__pyx_v_c->sent_adresses[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_10); - /* "average_inner.pyx":184 + /* "average_inner.pyx":203 * for token in obj[0]: * c.sent_adresses[eff_words] = obj[1] * if token in vocab: # <<<<<<<<<<<<<< * # In Vocabulary * word = vocab[token] */ - __pyx_t_6 = (__Pyx_PySequence_ContainsTF(__pyx_v_token, __pyx_v_vocab, Py_EQ)); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(0, 184, __pyx_L1_error) + __pyx_t_6 = (__Pyx_PySequence_ContainsTF(__pyx_v_token, __pyx_v_vocab, Py_EQ)); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(0, 203, __pyx_L1_error) __pyx_t_5 = (__pyx_t_6 != 0); if (__pyx_t_5) { - /* "average_inner.pyx":186 + /* "average_inner.pyx":205 * if token in vocab: * # In Vocabulary * word = vocab[token] # <<<<<<<<<<<<<< * c.word_indices[eff_words] = word.index * c.subwords_idx_len[eff_words] = ZERO */ - __pyx_t_4 = __Pyx_PyObject_GetItem(__pyx_v_vocab, __pyx_v_token); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 186, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetItem(__pyx_v_vocab, __pyx_v_token); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 205, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_XDECREF_SET(__pyx_v_word, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":187 + /* "average_inner.pyx":206 * # In Vocabulary * word = vocab[token] * c.word_indices[eff_words] = word.index # <<<<<<<<<<<<<< * c.subwords_idx_len[eff_words] = ZERO * else: */ - __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_v_word, __pyx_n_s_index); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 187, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_v_word, __pyx_n_s_index); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 206, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 187, __pyx_L1_error) + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 206, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; (__pyx_v_c->word_indices[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_10); - /* "average_inner.pyx":188 + /* "average_inner.pyx":207 * word = vocab[token] * c.word_indices[eff_words] = word.index * c.subwords_idx_len[eff_words] = ZERO # <<<<<<<<<<<<<< @@ -3052,7 +3052,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ (__pyx_v_c->subwords_idx_len[__pyx_v_eff_words]) = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":184 + /* "average_inner.pyx":203 * for token in obj[0]: * c.sent_adresses[eff_words] = obj[1] * if token in vocab: # <<<<<<<<<<<<<< @@ -3062,7 +3062,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 goto __pyx_L8; } - /* "average_inner.pyx":191 + /* "average_inner.pyx":210 * else: * # OOV words --> write ngram indices to memory * c.word_indices[eff_words] = ZERO # <<<<<<<<<<<<<< @@ -3072,18 +3072,18 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 /*else*/ { (__pyx_v_c->word_indices[__pyx_v_eff_words]) = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":193 + /* "average_inner.pyx":212 * c.word_indices[eff_words] = ZERO * * encoded_ngrams = compute_ngrams_bytes(token, c.min_n, c.max_n) # <<<<<<<<<<<<<< * hashes = [ft_hash_bytes(n) % c.bucket for n in encoded_ngrams] * */ - __Pyx_GetModuleGlobalName(__pyx_t_11, __pyx_n_s_compute_ngrams_bytes); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 193, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_11, __pyx_n_s_compute_ngrams_bytes); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 212, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_11); - __pyx_t_12 = __Pyx_PyInt_From_int(__pyx_v_c->min_n); if (unlikely(!__pyx_t_12)) __PYX_ERR(0, 193, __pyx_L1_error) + __pyx_t_12 = __Pyx_PyInt_From_int(__pyx_v_c->min_n); if (unlikely(!__pyx_t_12)) __PYX_ERR(0, 212, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_12); - __pyx_t_13 = __Pyx_PyInt_From_int(__pyx_v_c->max_n); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 193, __pyx_L1_error) + __pyx_t_13 = __Pyx_PyInt_From_int(__pyx_v_c->max_n); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 212, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_13); __pyx_t_14 = NULL; __pyx_t_15 = 0; @@ -3100,7 +3100,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 #if CYTHON_FAST_PYCALL if (PyFunction_Check(__pyx_t_11)) { PyObject *__pyx_temp[4] = {__pyx_t_14, __pyx_v_token, __pyx_t_12, __pyx_t_13}; - __pyx_t_4 = __Pyx_PyFunction_FastCall(__pyx_t_11, __pyx_temp+1-__pyx_t_15, 3+__pyx_t_15); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 193, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyFunction_FastCall(__pyx_t_11, __pyx_temp+1-__pyx_t_15, 3+__pyx_t_15); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 212, __pyx_L1_error) __Pyx_XDECREF(__pyx_t_14); __pyx_t_14 = 0; __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0; @@ -3110,7 +3110,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 #if CYTHON_FAST_PYCCALL if (__Pyx_PyFastCFunction_Check(__pyx_t_11)) { PyObject *__pyx_temp[4] = {__pyx_t_14, __pyx_v_token, __pyx_t_12, __pyx_t_13}; - __pyx_t_4 = __Pyx_PyCFunction_FastCall(__pyx_t_11, __pyx_temp+1-__pyx_t_15, 3+__pyx_t_15); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 193, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyCFunction_FastCall(__pyx_t_11, __pyx_temp+1-__pyx_t_15, 3+__pyx_t_15); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 212, __pyx_L1_error) __Pyx_XDECREF(__pyx_t_14); __pyx_t_14 = 0; __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0; @@ -3118,7 +3118,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 } else #endif { - __pyx_t_16 = PyTuple_New(3+__pyx_t_15); if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 193, __pyx_L1_error) + __pyx_t_16 = PyTuple_New(3+__pyx_t_15); if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 212, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_16); if (__pyx_t_14) { __Pyx_GIVEREF(__pyx_t_14); PyTuple_SET_ITEM(__pyx_t_16, 0, __pyx_t_14); __pyx_t_14 = NULL; @@ -3132,7 +3132,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 PyTuple_SET_ITEM(__pyx_t_16, 2+__pyx_t_15, __pyx_t_13); __pyx_t_12 = 0; __pyx_t_13 = 0; - __pyx_t_4 = __Pyx_PyObject_Call(__pyx_t_11, __pyx_t_16, NULL); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 193, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_Call(__pyx_t_11, __pyx_t_16, NULL); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 212, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_16); __pyx_t_16 = 0; } @@ -3140,39 +3140,39 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __Pyx_XDECREF_SET(__pyx_v_encoded_ngrams, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":194 + /* "average_inner.pyx":213 * * encoded_ngrams = compute_ngrams_bytes(token, c.min_n, c.max_n) * hashes = [ft_hash_bytes(n) % c.bucket for n in encoded_ngrams] # <<<<<<<<<<<<<< * * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) */ - __pyx_t_4 = PyList_New(0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 194, __pyx_L1_error) + __pyx_t_4 = PyList_New(0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); if (likely(PyList_CheckExact(__pyx_v_encoded_ngrams)) || PyTuple_CheckExact(__pyx_v_encoded_ngrams)) { __pyx_t_11 = __pyx_v_encoded_ngrams; __Pyx_INCREF(__pyx_t_11); __pyx_t_17 = 0; __pyx_t_18 = NULL; } else { - __pyx_t_17 = -1; __pyx_t_11 = PyObject_GetIter(__pyx_v_encoded_ngrams); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 194, __pyx_L1_error) + __pyx_t_17 = -1; __pyx_t_11 = PyObject_GetIter(__pyx_v_encoded_ngrams); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_11); - __pyx_t_18 = Py_TYPE(__pyx_t_11)->tp_iternext; if (unlikely(!__pyx_t_18)) __PYX_ERR(0, 194, __pyx_L1_error) + __pyx_t_18 = Py_TYPE(__pyx_t_11)->tp_iternext; if (unlikely(!__pyx_t_18)) __PYX_ERR(0, 213, __pyx_L1_error) } for (;;) { if (likely(!__pyx_t_18)) { if (likely(PyList_CheckExact(__pyx_t_11))) { if (__pyx_t_17 >= PyList_GET_SIZE(__pyx_t_11)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_16 = PyList_GET_ITEM(__pyx_t_11, __pyx_t_17); __Pyx_INCREF(__pyx_t_16); __pyx_t_17++; if (unlikely(0 < 0)) __PYX_ERR(0, 194, __pyx_L1_error) + __pyx_t_16 = PyList_GET_ITEM(__pyx_t_11, __pyx_t_17); __Pyx_INCREF(__pyx_t_16); __pyx_t_17++; if (unlikely(0 < 0)) __PYX_ERR(0, 213, __pyx_L1_error) #else - __pyx_t_16 = PySequence_ITEM(__pyx_t_11, __pyx_t_17); __pyx_t_17++; if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 194, __pyx_L1_error) + __pyx_t_16 = PySequence_ITEM(__pyx_t_11, __pyx_t_17); __pyx_t_17++; if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_16); #endif } else { if (__pyx_t_17 >= PyTuple_GET_SIZE(__pyx_t_11)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_16 = PyTuple_GET_ITEM(__pyx_t_11, __pyx_t_17); __Pyx_INCREF(__pyx_t_16); __pyx_t_17++; if (unlikely(0 < 0)) __PYX_ERR(0, 194, __pyx_L1_error) + __pyx_t_16 = PyTuple_GET_ITEM(__pyx_t_11, __pyx_t_17); __Pyx_INCREF(__pyx_t_16); __pyx_t_17++; if (unlikely(0 < 0)) __PYX_ERR(0, 213, __pyx_L1_error) #else - __pyx_t_16 = PySequence_ITEM(__pyx_t_11, __pyx_t_17); __pyx_t_17++; if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 194, __pyx_L1_error) + __pyx_t_16 = PySequence_ITEM(__pyx_t_11, __pyx_t_17); __pyx_t_17++; if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_16); #endif } @@ -3182,7 +3182,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 PyObject* exc_type = PyErr_Occurred(); if (exc_type) { if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); - else __PYX_ERR(0, 194, __pyx_L1_error) + else __PYX_ERR(0, 213, __pyx_L1_error) } break; } @@ -3190,7 +3190,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 } __Pyx_XDECREF_SET(__pyx_v_n, __pyx_t_16); __pyx_t_16 = 0; - __Pyx_GetModuleGlobalName(__pyx_t_13, __pyx_n_s_ft_hash_bytes); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 194, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_13, __pyx_n_s_ft_hash_bytes); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_13); __pyx_t_12 = NULL; if (CYTHON_UNPACK_METHODS && unlikely(PyMethod_Check(__pyx_t_13))) { @@ -3204,23 +3204,23 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 } __pyx_t_16 = (__pyx_t_12) ? __Pyx_PyObject_Call2Args(__pyx_t_13, __pyx_t_12, __pyx_v_n) : __Pyx_PyObject_CallOneArg(__pyx_t_13, __pyx_v_n); __Pyx_XDECREF(__pyx_t_12); __pyx_t_12 = 0; - if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 194, __pyx_L1_error) + if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_16); __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0; - __pyx_t_13 = __Pyx_PyInt_From_int(__pyx_v_c->bucket); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 194, __pyx_L1_error) + __pyx_t_13 = __Pyx_PyInt_From_int(__pyx_v_c->bucket); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_13); - __pyx_t_12 = PyNumber_Remainder(__pyx_t_16, __pyx_t_13); if (unlikely(!__pyx_t_12)) __PYX_ERR(0, 194, __pyx_L1_error) + __pyx_t_12 = PyNumber_Remainder(__pyx_t_16, __pyx_t_13); if (unlikely(!__pyx_t_12)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_12); __Pyx_DECREF(__pyx_t_16); __pyx_t_16 = 0; __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0; - if (unlikely(__Pyx_ListComp_Append(__pyx_t_4, (PyObject*)__pyx_t_12))) __PYX_ERR(0, 194, __pyx_L1_error) + if (unlikely(__Pyx_ListComp_Append(__pyx_t_4, (PyObject*)__pyx_t_12))) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0; } __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 0; __Pyx_XDECREF_SET(__pyx_v_hashes, ((PyObject*)__pyx_t_4)); __pyx_t_4 = 0; - /* "average_inner.pyx":196 + /* "average_inner.pyx":215 * hashes = [ft_hash_bytes(n) % c.bucket for n in encoded_ngrams] * * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) # <<<<<<<<<<<<<< @@ -3228,7 +3228,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 * c.subwords_idx[(eff_words * MAX_NGRAMS) + i] = h */ __pyx_t_19 = 40; - __pyx_t_17 = PyObject_Length(__pyx_v_encoded_ngrams); if (unlikely(__pyx_t_17 == ((Py_ssize_t)-1))) __PYX_ERR(0, 196, __pyx_L1_error) + __pyx_t_17 = PyObject_Length(__pyx_v_encoded_ngrams); if (unlikely(__pyx_t_17 == ((Py_ssize_t)-1))) __PYX_ERR(0, 215, __pyx_L1_error) if (((__pyx_t_19 < __pyx_t_17) != 0)) { __pyx_t_20 = __pyx_t_19; } else { @@ -3236,7 +3236,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 } (__pyx_v_c->subwords_idx_len[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_20); - /* "average_inner.pyx":197 + /* "average_inner.pyx":216 * * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) * for i, h in enumerate(hashes[:MAX_NGRAMS]): # <<<<<<<<<<<<<< @@ -3245,46 +3245,46 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ __Pyx_INCREF(__pyx_int_0); __pyx_t_4 = __pyx_int_0; - __pyx_t_11 = __Pyx_PyList_GetSlice(__pyx_v_hashes, 0, 40); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 197, __pyx_L1_error) + __pyx_t_11 = __Pyx_PyList_GetSlice(__pyx_v_hashes, 0, 40); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 216, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_11); __pyx_t_12 = __pyx_t_11; __Pyx_INCREF(__pyx_t_12); __pyx_t_20 = 0; __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 0; for (;;) { if (__pyx_t_20 >= PyList_GET_SIZE(__pyx_t_12)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_11 = PyList_GET_ITEM(__pyx_t_12, __pyx_t_20); __Pyx_INCREF(__pyx_t_11); __pyx_t_20++; if (unlikely(0 < 0)) __PYX_ERR(0, 197, __pyx_L1_error) + __pyx_t_11 = PyList_GET_ITEM(__pyx_t_12, __pyx_t_20); __Pyx_INCREF(__pyx_t_11); __pyx_t_20++; if (unlikely(0 < 0)) __PYX_ERR(0, 216, __pyx_L1_error) #else - __pyx_t_11 = PySequence_ITEM(__pyx_t_12, __pyx_t_20); __pyx_t_20++; if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 197, __pyx_L1_error) + __pyx_t_11 = PySequence_ITEM(__pyx_t_12, __pyx_t_20); __pyx_t_20++; if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 216, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_11); #endif __Pyx_XDECREF_SET(__pyx_v_h, __pyx_t_11); __pyx_t_11 = 0; __Pyx_INCREF(__pyx_t_4); __Pyx_XDECREF_SET(__pyx_v_i, __pyx_t_4); - __pyx_t_11 = __Pyx_PyInt_AddObjC(__pyx_t_4, __pyx_int_1, 1, 0, 0); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 197, __pyx_L1_error) + __pyx_t_11 = __Pyx_PyInt_AddObjC(__pyx_t_4, __pyx_int_1, 1, 0, 0); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 216, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_11); __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = __pyx_t_11; __pyx_t_11 = 0; - /* "average_inner.pyx":198 + /* "average_inner.pyx":217 * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) * for i, h in enumerate(hashes[:MAX_NGRAMS]): * c.subwords_idx[(eff_words * MAX_NGRAMS) + i] = h # <<<<<<<<<<<<<< * * eff_words += ONE */ - __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_v_h); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 198, __pyx_L1_error) - __pyx_t_11 = __Pyx_PyInt_From_long((__pyx_v_eff_words * 40)); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 198, __pyx_L1_error) + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_v_h); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 217, __pyx_L1_error) + __pyx_t_11 = __Pyx_PyInt_From_long((__pyx_v_eff_words * 40)); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 217, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_11); - __pyx_t_13 = PyNumber_Add(__pyx_t_11, __pyx_v_i); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 198, __pyx_L1_error) + __pyx_t_13 = PyNumber_Add(__pyx_t_11, __pyx_v_i); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 217, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_13); __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 0; - __pyx_t_17 = __Pyx_PyIndex_AsSsize_t(__pyx_t_13); if (unlikely((__pyx_t_17 == (Py_ssize_t)-1) && PyErr_Occurred())) __PYX_ERR(0, 198, __pyx_L1_error) + __pyx_t_17 = __Pyx_PyIndex_AsSsize_t(__pyx_t_13); if (unlikely((__pyx_t_17 == (Py_ssize_t)-1) && PyErr_Occurred())) __PYX_ERR(0, 217, __pyx_L1_error) __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0; (__pyx_v_c->subwords_idx[__pyx_t_17]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_10); - /* "average_inner.pyx":197 + /* "average_inner.pyx":216 * * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) * for i, h in enumerate(hashes[:MAX_NGRAMS]): # <<<<<<<<<<<<<< @@ -3297,7 +3297,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 } __pyx_L8:; - /* "average_inner.pyx":200 + /* "average_inner.pyx":219 * c.subwords_idx[(eff_words * MAX_NGRAMS) + i] = h * * eff_words += ONE # <<<<<<<<<<<<<< @@ -3306,7 +3306,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ __pyx_v_eff_words = (__pyx_v_eff_words + __pyx_v_13average_inner_ONE); - /* "average_inner.pyx":202 + /* "average_inner.pyx":221 * eff_words += ONE * * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -3316,7 +3316,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __pyx_t_5 = ((__pyx_v_eff_words == 0x2710) != 0); if (__pyx_t_5) { - /* "average_inner.pyx":203 + /* "average_inner.pyx":222 * * if eff_words == MAX_WORDS: * break # <<<<<<<<<<<<<< @@ -3325,7 +3325,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ goto __pyx_L7_break; - /* "average_inner.pyx":202 + /* "average_inner.pyx":221 * eff_words += ONE * * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -3334,7 +3334,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ } - /* "average_inner.pyx":182 + /* "average_inner.pyx":201 * if not obj[0]: * continue * for token in obj[0]: # <<<<<<<<<<<<<< @@ -3345,7 +3345,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __pyx_L7_break:; __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; - /* "average_inner.pyx":205 + /* "average_inner.pyx":224 * break * * eff_sents += 1 # <<<<<<<<<<<<<< @@ -3354,7 +3354,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ __pyx_v_eff_sents = (__pyx_v_eff_sents + 1); - /* "average_inner.pyx":206 + /* "average_inner.pyx":225 * * eff_sents += 1 * c.sentence_boundary[eff_sents] = eff_words # <<<<<<<<<<<<<< @@ -3363,7 +3363,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ (__pyx_v_c->sentence_boundary[__pyx_v_eff_sents]) = __pyx_v_eff_words; - /* "average_inner.pyx":208 + /* "average_inner.pyx":227 * c.sentence_boundary[eff_sents] = eff_words * * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -3373,7 +3373,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __pyx_t_5 = ((__pyx_v_eff_words == 0x2710) != 0); if (__pyx_t_5) { - /* "average_inner.pyx":209 + /* "average_inner.pyx":228 * * if eff_words == MAX_WORDS: * break # <<<<<<<<<<<<<< @@ -3382,7 +3382,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ goto __pyx_L4_break; - /* "average_inner.pyx":208 + /* "average_inner.pyx":227 * c.sentence_boundary[eff_sents] = eff_words * * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -3391,7 +3391,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ } - /* "average_inner.pyx":179 + /* "average_inner.pyx":198 * c.sentence_boundary[0] = ZERO * * for obj in indexed_sentences: # <<<<<<<<<<<<<< @@ -3403,19 +3403,19 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __pyx_L4_break:; __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":211 + /* "average_inner.pyx":230 * break * * return eff_sents, eff_words # <<<<<<<<<<<<<< * - * cdef void compute_base_sentence_averages(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: + * cdef void compute_base_sentence_averages( */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sents); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 211, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sents); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 230, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_7 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 211, __pyx_L1_error) + __pyx_t_7 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 230, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); - __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 211, __pyx_L1_error) + __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 230, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_1); @@ -3427,12 +3427,12 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __pyx_t_4 = 0; goto __pyx_L0; - /* "average_inner.pyx":150 + /* "average_inner.pyx":165 * return eff_sents, eff_words * - * cdef object populate_ft_s2v_config(FTSentenceVecsConfig *c, vocab, indexed_sentences): # <<<<<<<<<<<<<< - * """Prepare C structures for FastText so we can go "full C" and release the Python GIL. - * + * cdef object populate_ft_s2v_config( # <<<<<<<<<<<<<< + * FTSentenceVecsConfig *c, + * vocab, */ /* function exit code */ @@ -3461,12 +3461,12 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 return __pyx_r; } -/* "average_inner.pyx":213 +/* "average_inner.pyx":232 * return eff_sents, eff_words * - * cdef void compute_base_sentence_averages(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< - * """Perform optimized sentence-level averaging for BaseAny2Vec model. - * + * cdef void compute_base_sentence_averages( # <<<<<<<<<<<<<< + * BaseSentenceVecsConfig *c, + * uINT_t num_sentences, */ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { @@ -3489,7 +3489,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_t_13average_inner_uINT_t __pyx_t_7; int __pyx_t_8; - /* "average_inner.pyx":229 + /* "average_inner.pyx":251 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -3499,7 +3499,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "average_inner.pyx":237 + /* "average_inner.pyx":259 * REAL_t sent_len, inv_count * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -3511,7 +3511,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) { __pyx_v_sent_idx = __pyx_t_4; - /* "average_inner.pyx":238 + /* "average_inner.pyx":260 * * for sent_idx in range(num_sentences): * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -3520,7 +3520,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "average_inner.pyx":240 + /* "average_inner.pyx":262 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -3529,7 +3529,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "average_inner.pyx":241 + /* "average_inner.pyx":263 * * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -3538,7 +3538,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "average_inner.pyx":242 + /* "average_inner.pyx":264 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -3547,7 +3547,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "average_inner.pyx":244 + /* "average_inner.pyx":266 * sent_len = ZEROF * * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -3559,7 +3559,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { __pyx_v_i = __pyx_t_7; - /* "average_inner.pyx":245 + /* "average_inner.pyx":267 * * for i in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -3568,7 +3568,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "average_inner.pyx":246 + /* "average_inner.pyx":268 * for i in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< @@ -3577,7 +3577,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); - /* "average_inner.pyx":247 + /* "average_inner.pyx":269 * sent_len += ONEF * sent_row = c.sent_adresses[i] * size * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< @@ -3586,7 +3586,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); - /* "average_inner.pyx":248 + /* "average_inner.pyx":270 * sent_row = c.sent_adresses[i] * size * word_row = c.word_indices[i] * size * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< @@ -3595,7 +3595,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); - /* "average_inner.pyx":250 + /* "average_inner.pyx":272 * word_idx = c.word_indices[i] * * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) # <<<<<<<<<<<<<< @@ -3605,7 +3605,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); } - /* "average_inner.pyx":252 + /* "average_inner.pyx":274 * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< @@ -3615,7 +3615,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_t_8 = ((__pyx_v_sent_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_8) { - /* "average_inner.pyx":253 + /* "average_inner.pyx":275 * * if sent_len > ZEROF: * inv_count = ONEF / sent_len # <<<<<<<<<<<<<< @@ -3624,16 +3624,16 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_sent_len); - /* "average_inner.pyx":256 + /* "average_inner.pyx":278 * # If we perform the a*x on memory, the computation is compatible with many-to-one mappings * # because it doesn't rescale the overall result * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) # <<<<<<<<<<<<<< * - * cdef void compute_ft_sentence_averages(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: + * cdef void compute_ft_sentence_averages( */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&__pyx_v_13average_inner_ONE)); - /* "average_inner.pyx":252 + /* "average_inner.pyx":274 * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< @@ -3643,23 +3643,23 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ } } - /* "average_inner.pyx":213 + /* "average_inner.pyx":232 * return eff_sents, eff_words * - * cdef void compute_base_sentence_averages(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< - * """Perform optimized sentence-level averaging for BaseAny2Vec model. - * + * cdef void compute_base_sentence_averages( # <<<<<<<<<<<<<< + * BaseSentenceVecsConfig *c, + * uINT_t num_sentences, */ /* function exit code */ } -/* "average_inner.pyx":258 +/* "average_inner.pyx":280 * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) * - * cdef void compute_ft_sentence_averages(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< - * """Perform optimized sentence-level averaging for FastText model. - * + * cdef void compute_ft_sentence_averages( # <<<<<<<<<<<<<< + * FTSentenceVecsConfig *c, + * uINT_t num_sentences, */ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { @@ -3691,7 +3691,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_13average_inner_uINT_t __pyx_t_11; __pyx_t_13average_inner_uINT_t __pyx_t_12; - /* "average_inner.pyx":274 + /* "average_inner.pyx":299 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -3701,7 +3701,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "average_inner.pyx":284 + /* "average_inner.pyx":309 * REAL_t sent_len * REAL_t inv_count, inv_ngram * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< @@ -3711,7 +3711,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_2 = __pyx_v_c->oov_weight; __pyx_v_oov_weight = __pyx_t_2; - /* "average_inner.pyx":287 + /* "average_inner.pyx":312 * * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -3723,7 +3723,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { __pyx_v_sent_idx = __pyx_t_5; - /* "average_inner.pyx":288 + /* "average_inner.pyx":313 * * for sent_idx in range(num_sentences): * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -3732,7 +3732,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "average_inner.pyx":289 + /* "average_inner.pyx":314 * for sent_idx in range(num_sentences): * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -3741,7 +3741,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "average_inner.pyx":290 + /* "average_inner.pyx":315 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -3750,7 +3750,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "average_inner.pyx":291 + /* "average_inner.pyx":316 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -3759,7 +3759,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "average_inner.pyx":293 + /* "average_inner.pyx":318 * sent_len = ZEROF * * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -3771,7 +3771,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { __pyx_v_i = __pyx_t_8; - /* "average_inner.pyx":294 + /* "average_inner.pyx":319 * * for i in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -3780,7 +3780,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "average_inner.pyx":295 + /* "average_inner.pyx":320 * for i in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< @@ -3789,7 +3789,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); - /* "average_inner.pyx":297 + /* "average_inner.pyx":322 * sent_row = c.sent_adresses[i] * size * * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< @@ -3798,7 +3798,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); - /* "average_inner.pyx":298 + /* "average_inner.pyx":323 * * word_idx = c.word_indices[i] * ngrams = c.subwords_idx_len[i] # <<<<<<<<<<<<<< @@ -3807,7 +3807,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_i]); - /* "average_inner.pyx":300 + /* "average_inner.pyx":325 * ngrams = c.subwords_idx_len[i] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -3817,7 +3817,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); if (__pyx_t_9) { - /* "average_inner.pyx":301 + /* "average_inner.pyx":326 * * if ngrams == 0: * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< @@ -3826,7 +3826,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); - /* "average_inner.pyx":302 + /* "average_inner.pyx":327 * if ngrams == 0: * word_row = c.word_indices[i] * size * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) # <<<<<<<<<<<<<< @@ -3835,7 +3835,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); - /* "average_inner.pyx":300 + /* "average_inner.pyx":325 * ngrams = c.subwords_idx_len[i] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -3845,7 +3845,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ goto __pyx_L7; } - /* "average_inner.pyx":304 + /* "average_inner.pyx":329 * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) * else: * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< @@ -3855,7 +3855,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ /*else*/ { __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); - /* "average_inner.pyx":305 + /* "average_inner.pyx":330 * else: * inv_ngram = (ONEF / ngrams) * c.oov_weight * for j in range(ngrams): # <<<<<<<<<<<<<< @@ -3867,7 +3867,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ for (__pyx_t_12 = 0; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { __pyx_v_j = __pyx_t_12; - /* "average_inner.pyx":306 + /* "average_inner.pyx":331 * inv_ngram = (ONEF / ngrams) * c.oov_weight * for j in range(ngrams): * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size # <<<<<<<<<<<<<< @@ -3876,7 +3876,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_i * 40) + __pyx_v_j)]) * __pyx_v_size); - /* "average_inner.pyx":307 + /* "average_inner.pyx":332 * for j in range(ngrams): * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) # <<<<<<<<<<<<<< @@ -3889,7 +3889,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_L7:; } - /* "average_inner.pyx":309 + /* "average_inner.pyx":334 * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< @@ -3899,7 +3899,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_9 = ((__pyx_v_sent_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_9) { - /* "average_inner.pyx":310 + /* "average_inner.pyx":335 * * if sent_len > ZEROF: * inv_count = ONEF / sent_len # <<<<<<<<<<<<<< @@ -3908,16 +3908,16 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_sent_len); - /* "average_inner.pyx":311 + /* "average_inner.pyx":336 * if sent_len > ZEROF: * inv_count = ONEF / sent_len * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) # <<<<<<<<<<<<<< * - * def train_average_cy(model, indexed_sentences, target, memory): + * def train_average_cy( */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&__pyx_v_13average_inner_ONE)); - /* "average_inner.pyx":309 + /* "average_inner.pyx":334 * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< @@ -3927,23 +3927,23 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ } } - /* "average_inner.pyx":258 + /* "average_inner.pyx":280 * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) * - * cdef void compute_ft_sentence_averages(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: # <<<<<<<<<<<<<< - * """Perform optimized sentence-level averaging for FastText model. - * + * cdef void compute_ft_sentence_averages( # <<<<<<<<<<<<<< + * FTSentenceVecsConfig *c, + * uINT_t num_sentences, */ /* function exit code */ } -/* "average_inner.pyx":313 +/* "average_inner.pyx":338 * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) * - * def train_average_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< - * """Training on a sequence of sentences and update the target ndarray. - * + * def train_average_cy( # <<<<<<<<<<<<<< + * model, + * indexed_sentences, */ /* Python wrapper */ @@ -3985,23 +3985,23 @@ static PyObject *__pyx_pw_13average_inner_1train_average_cy(PyObject *__pyx_self case 1: if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 1); __PYX_ERR(0, 313, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 1); __PYX_ERR(0, 338, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 2: if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 2); __PYX_ERR(0, 313, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 2); __PYX_ERR(0, 338, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 3: if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 3); __PYX_ERR(0, 313, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 3); __PYX_ERR(0, 338, __pyx_L3_error) } } if (unlikely(kw_args > 0)) { - if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_average_cy") < 0)) __PYX_ERR(0, 313, __pyx_L3_error) + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_average_cy") < 0)) __PYX_ERR(0, 338, __pyx_L3_error) } } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { goto __pyx_L5_argtuple_error; @@ -4018,7 +4018,7 @@ static PyObject *__pyx_pw_13average_inner_1train_average_cy(PyObject *__pyx_self } goto __pyx_L4_argument_unpacking_done; __pyx_L5_argtuple_error:; - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 313, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 338, __pyx_L3_error) __pyx_L3_error:; __Pyx_AddTraceback("average_inner.train_average_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __Pyx_RefNannyFinishContext(); @@ -4049,7 +4049,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_t_13average_inner_uINT_t __pyx_t_9; __Pyx_RefNannySetupContext("train_average_cy", 0); - /* "average_inner.pyx":337 + /* "average_inner.pyx":367 * """ * * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< @@ -4058,7 +4058,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_sentences = 0; - /* "average_inner.pyx":338 + /* "average_inner.pyx":368 * * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< @@ -4067,44 +4067,52 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_words = 0; - /* "average_inner.pyx":342 + /* "average_inner.pyx":372 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< * init_base_s2v_config(&w2v, model, target, memory) * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 342, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 372, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 342, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 372, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_3 = ((!__pyx_t_2) != 0); if (__pyx_t_3) { - /* "average_inner.pyx":343 + /* "average_inner.pyx":373 * * if not model.is_ft: * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< * - * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + * eff_sentences, eff_words = populate_base_s2v_config( */ - __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 343, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 373, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":345 - * init_base_s2v_config(&w2v, model, target, memory) - * - * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) # <<<<<<<<<<<<<< - * - * with nogil: + /* "average_inner.pyx":377 + * eff_sentences, eff_words = populate_base_s2v_config( + * &w2v, + * model.wv.vocab, # <<<<<<<<<<<<<< + * indexed_sentences, + * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 345, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 377, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 345, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 377, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 345, __pyx_L1_error) + + /* "average_inner.pyx":375 + * init_base_s2v_config(&w2v, model, target, memory) + * + * eff_sentences, eff_words = populate_base_s2v_config( # <<<<<<<<<<<<<< + * &w2v, + * model.wv.vocab, + */ + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 375, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -4113,7 +4121,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 345, __pyx_L1_error) + __PYX_ERR(0, 375, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -4126,15 +4134,15 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_4); __Pyx_INCREF(__pyx_t_5); #else - __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 345, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 375, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 345, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 375, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 345, __pyx_L1_error) + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 375, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; @@ -4142,7 +4150,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_4); index = 1; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L4_unpacking_failed; __Pyx_GOTREF(__pyx_t_5); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 345, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 375, __pyx_L1_error) __pyx_t_7 = NULL; __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; goto __pyx_L5_unpacking_done; @@ -4150,22 +4158,22 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_t_7 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 345, __pyx_L1_error) + __PYX_ERR(0, 375, __pyx_L1_error) __pyx_L5_unpacking_done:; } - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 345, __pyx_L1_error) + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 375, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 345, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 375, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; __pyx_v_eff_sentences = __pyx_t_8; __pyx_v_eff_words = __pyx_t_9; - /* "average_inner.pyx":347 - * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + /* "average_inner.pyx":381 + * ) * * with nogil: # <<<<<<<<<<<<<< - * compute_base_sentence_averages(&w2v, eff_sentences) - * else: + * compute_base_sentence_averages( + * &w2v, */ { #ifdef WITH_THREAD @@ -4175,22 +4183,22 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "average_inner.pyx":348 + /* "average_inner.pyx":382 * * with nogil: - * compute_base_sentence_averages(&w2v, eff_sentences) # <<<<<<<<<<<<<< - * else: - * init_ft_s2v_config(&ft, model, target, memory) + * compute_base_sentence_averages( # <<<<<<<<<<<<<< + * &w2v, + * eff_sentences, */ __pyx_f_13average_inner_compute_base_sentence_averages((&__pyx_v_w2v), __pyx_v_eff_sentences); } - /* "average_inner.pyx":347 - * eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + /* "average_inner.pyx":381 + * ) * * with nogil: # <<<<<<<<<<<<<< - * compute_base_sentence_averages(&w2v, eff_sentences) - * else: + * compute_base_sentence_averages( + * &w2v, */ /*finally:*/ { /*normal exit:*/{ @@ -4204,7 +4212,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec } } - /* "average_inner.pyx":342 + /* "average_inner.pyx":372 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< @@ -4214,31 +4222,47 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec goto __pyx_L3; } - /* "average_inner.pyx":350 - * compute_base_sentence_averages(&w2v, eff_sentences) + /* "average_inner.pyx":387 + * ) * else: - * init_ft_s2v_config(&ft, model, target, memory) # <<<<<<<<<<<<<< - * - * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + * init_ft_s2v_config( # <<<<<<<<<<<<<< + * &ft, + * model, */ /*else*/ { - __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 350, __pyx_L1_error) + + /* "average_inner.pyx":391 + * model, + * target, + * memory, # <<<<<<<<<<<<<< + * ) + * + */ + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 387, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":352 - * init_ft_s2v_config(&ft, model, target, memory) - * - * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) # <<<<<<<<<<<<<< - * - * with nogil: + /* "average_inner.pyx":396 + * eff_sentences, eff_words = populate_ft_s2v_config( + * &ft, + * model.wv.vocab, # <<<<<<<<<<<<<< + * indexed_sentences, + * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 352, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 396, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 352, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 396, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 352, __pyx_L1_error) + + /* "average_inner.pyx":394 + * ) + * + * eff_sentences, eff_words = populate_ft_s2v_config( # <<<<<<<<<<<<<< + * &ft, + * model.wv.vocab, + */ + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 394, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -4247,7 +4271,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 352, __pyx_L1_error) + __PYX_ERR(0, 394, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -4260,15 +4284,15 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_5); __Pyx_INCREF(__pyx_t_4); #else - __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 352, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 394, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); - __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 352, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 394, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 352, __pyx_L1_error) + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 394, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; @@ -4276,7 +4300,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_5); index = 1; __pyx_t_4 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_4)) goto __pyx_L9_unpacking_failed; __Pyx_GOTREF(__pyx_t_4); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 352, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 394, __pyx_L1_error) __pyx_t_7 = NULL; __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; goto __pyx_L10_unpacking_done; @@ -4284,22 +4308,22 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_t_7 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 352, __pyx_L1_error) + __PYX_ERR(0, 394, __pyx_L1_error) __pyx_L10_unpacking_done:; } - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 352, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 394, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 352, __pyx_L1_error) + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 394, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; __pyx_v_eff_sentences = __pyx_t_9; __pyx_v_eff_words = __pyx_t_8; - /* "average_inner.pyx":354 - * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + /* "average_inner.pyx":400 + * ) * * with nogil: # <<<<<<<<<<<<<< - * compute_ft_sentence_averages(&ft, eff_sentences) - * + * compute_ft_sentence_averages( + * &ft, */ { #ifdef WITH_THREAD @@ -4309,22 +4333,22 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "average_inner.pyx":355 + /* "average_inner.pyx":401 * * with nogil: - * compute_ft_sentence_averages(&ft, eff_sentences) # <<<<<<<<<<<<<< - * - * return eff_sentences, eff_words + * compute_ft_sentence_averages( # <<<<<<<<<<<<<< + * &ft, + * eff_sentences */ __pyx_f_13average_inner_compute_ft_sentence_averages((&__pyx_v_ft), __pyx_v_eff_sentences); } - /* "average_inner.pyx":354 - * eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + /* "average_inner.pyx":400 + * ) * * with nogil: # <<<<<<<<<<<<<< - * compute_ft_sentence_averages(&ft, eff_sentences) - * + * compute_ft_sentence_averages( + * &ft, */ /*finally:*/ { /*normal exit:*/{ @@ -4340,19 +4364,19 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec } __pyx_L3:; - /* "average_inner.pyx":357 - * compute_ft_sentence_averages(&ft, eff_sentences) + /* "average_inner.pyx":406 + * ) * * return eff_sentences, eff_words # <<<<<<<<<<<<<< * * def init(): */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 357, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 406, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 357, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 406, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 357, __pyx_L1_error) + __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 406, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_1); @@ -4364,12 +4388,12 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_t_5 = 0; goto __pyx_L0; - /* "average_inner.pyx":313 + /* "average_inner.pyx":338 * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) * - * def train_average_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< - * """Training on a sequence of sentences and update the target ndarray. - * + * def train_average_cy( # <<<<<<<<<<<<<< + * model, + * indexed_sentences, */ /* function exit code */ @@ -4386,7 +4410,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec return __pyx_r; } -/* "average_inner.pyx":359 +/* "average_inner.pyx":408 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -4414,7 +4438,7 @@ static PyObject *__pyx_pf_13average_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("init", 0); - /* "average_inner.pyx":360 + /* "average_inner.pyx":409 * * def init(): * return 1 # <<<<<<<<<<<<<< @@ -4426,7 +4450,7 @@ static PyObject *__pyx_pf_13average_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __pyx_r = __pyx_int_1; goto __pyx_L0; - /* "average_inner.pyx":359 + /* "average_inner.pyx":408 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -6973,8 +6997,8 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = { {0, 0, 0, 0, 0, 0, 0} }; static CYTHON_SMALL_CODE int __Pyx_InitCachedBuiltins(void) { - __pyx_builtin_enumerate = __Pyx_GetBuiltinName(__pyx_n_s_enumerate); if (!__pyx_builtin_enumerate) __PYX_ERR(0, 197, __pyx_L1_error) - __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 237, __pyx_L1_error) + __pyx_builtin_enumerate = __Pyx_GetBuiltinName(__pyx_n_s_enumerate); if (!__pyx_builtin_enumerate) __PYX_ERR(0, 216, __pyx_L1_error) + __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 259, __pyx_L1_error) __pyx_builtin_ValueError = __Pyx_GetBuiltinName(__pyx_n_s_ValueError); if (!__pyx_builtin_ValueError) __PYX_ERR(1, 272, __pyx_L1_error) __pyx_builtin_RuntimeError = __Pyx_GetBuiltinName(__pyx_n_s_RuntimeError); if (!__pyx_builtin_RuntimeError) __PYX_ERR(1, 856, __pyx_L1_error) __pyx_builtin_ImportError = __Pyx_GetBuiltinName(__pyx_n_s_ImportError); if (!__pyx_builtin_ImportError) __PYX_ERR(1, 1038, __pyx_L1_error) @@ -7064,26 +7088,26 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { __Pyx_GOTREF(__pyx_tuple__7); __Pyx_GIVEREF(__pyx_tuple__7); - /* "average_inner.pyx":313 + /* "average_inner.pyx":338 * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) * - * def train_average_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< - * """Training on a sequence of sentences and update the target ndarray. - * + * def train_average_cy( # <<<<<<<<<<<<<< + * model, + * indexed_sentences, */ - __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 313, __pyx_L1_error) + __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 338, __pyx_L1_error) __Pyx_GOTREF(__pyx_tuple__9); __Pyx_GIVEREF(__pyx_tuple__9); - __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_train_average_cy, 313, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 313, __pyx_L1_error) + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_train_average_cy, 338, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 338, __pyx_L1_error) - /* "average_inner.pyx":359 + /* "average_inner.pyx":408 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_init, 359, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 359, __pyx_L1_error) + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_init, 408, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 408, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -7409,26 +7433,26 @@ if (!__Pyx_RefNanny) { if (__Pyx_patch_abc() < 0) __PYX_ERR(0, 1, __pyx_L1_error) #endif - /* "average_inner.pyx":14 + /* "average_inner.pyx":15 * * import cython * import numpy as np # <<<<<<<<<<<<<< * * cimport numpy as np */ - __pyx_t_1 = __Pyx_Import(__pyx_n_s_numpy, 0, -1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 14, __pyx_L1_error) + __pyx_t_1 = __Pyx_Import(__pyx_n_s_numpy, 0, -1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 15, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_np, __pyx_t_1) < 0) __PYX_ERR(0, 14, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_np, __pyx_t_1) < 0) __PYX_ERR(0, 15, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":18 + /* "average_inner.pyx":19 * cimport numpy as np * * from gensim.models._utils_any2vec import compute_ngrams_bytes, ft_hash_bytes # <<<<<<<<<<<<<< * * from libc.string cimport memset */ - __pyx_t_1 = PyList_New(2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 18, __pyx_L1_error) + __pyx_t_1 = PyList_New(2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 19, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_INCREF(__pyx_n_s_compute_ngrams_bytes); __Pyx_GIVEREF(__pyx_n_s_compute_ngrams_bytes); @@ -7436,74 +7460,74 @@ if (!__Pyx_RefNanny) { __Pyx_INCREF(__pyx_n_s_ft_hash_bytes); __Pyx_GIVEREF(__pyx_n_s_ft_hash_bytes); PyList_SET_ITEM(__pyx_t_1, 1, __pyx_n_s_ft_hash_bytes); - __pyx_t_2 = __Pyx_Import(__pyx_n_s_gensim_models__utils_any2vec, __pyx_t_1, -1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 18, __pyx_L1_error) + __pyx_t_2 = __Pyx_Import(__pyx_n_s_gensim_models__utils_any2vec, __pyx_t_1, -1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 19, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_1 = __Pyx_ImportFrom(__pyx_t_2, __pyx_n_s_compute_ngrams_bytes); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 18, __pyx_L1_error) + __pyx_t_1 = __Pyx_ImportFrom(__pyx_t_2, __pyx_n_s_compute_ngrams_bytes); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 19, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_compute_ngrams_bytes, __pyx_t_1) < 0) __PYX_ERR(0, 18, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_compute_ngrams_bytes, __pyx_t_1) < 0) __PYX_ERR(0, 19, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_1 = __Pyx_ImportFrom(__pyx_t_2, __pyx_n_s_ft_hash_bytes); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 18, __pyx_L1_error) + __pyx_t_1 = __Pyx_ImportFrom(__pyx_t_2, __pyx_n_s_ft_hash_bytes); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 19, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_ft_hash_bytes, __pyx_t_1) < 0) __PYX_ERR(0, 18, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_ft_hash_bytes, __pyx_t_1) < 0) __PYX_ERR(0, 19, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "average_inner.pyx":22 + /* "average_inner.pyx":23 * from libc.string cimport memset * * import scipy.linalg.blas as fblas # <<<<<<<<<<<<<< * * cdef saxpy_ptr saxpy=PyCObject_AsVoidPtr(fblas.saxpy._cpointer) # y += alpha * x */ - __pyx_t_2 = PyList_New(1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 22, __pyx_L1_error) + __pyx_t_2 = PyList_New(1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 23, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_INCREF(__pyx_n_s__8); __Pyx_GIVEREF(__pyx_n_s__8); PyList_SET_ITEM(__pyx_t_2, 0, __pyx_n_s__8); - __pyx_t_1 = __Pyx_Import(__pyx_n_s_scipy_linalg_blas, __pyx_t_2, -1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 22, __pyx_L1_error) + __pyx_t_1 = __Pyx_Import(__pyx_n_s_scipy_linalg_blas, __pyx_t_2, -1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 23, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_1) < 0) __PYX_ERR(0, 22, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_1) < 0) __PYX_ERR(0, 23, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":24 + /* "average_inner.pyx":25 * import scipy.linalg.blas as fblas * * cdef saxpy_ptr saxpy=PyCObject_AsVoidPtr(fblas.saxpy._cpointer) # y += alpha * x # <<<<<<<<<<<<<< * cdef sscal_ptr sscal=PyCObject_AsVoidPtr(fblas.sscal._cpointer) # x = alpha * x * */ - __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_fblas); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 24, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_fblas); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 25, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_saxpy); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 24, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_saxpy); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 25, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_2, __pyx_n_s_cpointer); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 24, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_2, __pyx_n_s_cpointer); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 25, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; __pyx_v_13average_inner_saxpy = ((__pyx_t_13average_inner_saxpy_ptr)PyCObject_AsVoidPtr(__pyx_t_1)); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":25 + /* "average_inner.pyx":26 * * cdef saxpy_ptr saxpy=PyCObject_AsVoidPtr(fblas.saxpy._cpointer) # y += alpha * x * cdef sscal_ptr sscal=PyCObject_AsVoidPtr(fblas.sscal._cpointer) # x = alpha * x # <<<<<<<<<<<<<< * * cdef int ONE = 1 */ - __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_fblas); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 25, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_fblas); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 26, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_sscal); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 25, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_sscal); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 26, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_2, __pyx_n_s_cpointer); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 25, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_2, __pyx_n_s_cpointer); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 26, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; __pyx_v_13average_inner_sscal = ((__pyx_t_13average_inner_sscal_ptr)PyCObject_AsVoidPtr(__pyx_t_1)); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":27 + /* "average_inner.pyx":28 * cdef sscal_ptr sscal=PyCObject_AsVoidPtr(fblas.sscal._cpointer) # x = alpha * x * * cdef int ONE = 1 # <<<<<<<<<<<<<< @@ -7512,7 +7536,7 @@ if (!__Pyx_RefNanny) { */ __pyx_v_13average_inner_ONE = ((int)1); - /* "average_inner.pyx":28 + /* "average_inner.pyx":29 * * cdef int ONE = 1 * cdef int ZERO = 0 # <<<<<<<<<<<<<< @@ -7521,7 +7545,7 @@ if (!__Pyx_RefNanny) { */ __pyx_v_13average_inner_ZERO = ((int)0); - /* "average_inner.pyx":30 + /* "average_inner.pyx":31 * cdef int ZERO = 0 * * cdef REAL_t ONEF = 1.0 # <<<<<<<<<<<<<< @@ -7530,7 +7554,7 @@ if (!__Pyx_RefNanny) { */ __pyx_v_13average_inner_ONEF = ((__pyx_t_13average_inner_REAL_t)1.0); - /* "average_inner.pyx":31 + /* "average_inner.pyx":32 * * cdef REAL_t ONEF = 1.0 * cdef REAL_t ZEROF = 0.0 # <<<<<<<<<<<<<< @@ -7539,58 +7563,58 @@ if (!__Pyx_RefNanny) { */ __pyx_v_13average_inner_ZEROF = ((__pyx_t_13average_inner_REAL_t)0.0); - /* "average_inner.pyx":313 + /* "average_inner.pyx":338 * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) * - * def train_average_cy(model, indexed_sentences, target, memory): # <<<<<<<<<<<<<< - * """Training on a sequence of sentences and update the target ndarray. - * + * def train_average_cy( # <<<<<<<<<<<<<< + * model, + * indexed_sentences, */ - __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_1train_average_cy, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 313, __pyx_L1_error) + __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_1train_average_cy, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 338, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_average_cy, __pyx_t_1) < 0) __PYX_ERR(0, 313, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_average_cy, __pyx_t_1) < 0) __PYX_ERR(0, 338, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":359 + /* "average_inner.pyx":408 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_3init, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 359, __pyx_L1_error) + __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_3init, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 408, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_1) < 0) __PYX_ERR(0, 359, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_1) < 0) __PYX_ERR(0, 408, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":362 + /* "average_inner.pyx":411 * return 1 * * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 362, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 411, __pyx_L1_error) - /* "average_inner.pyx":363 + /* "average_inner.pyx":412 * * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 363, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 412, __pyx_L1_error) - /* "average_inner.pyx":364 + /* "average_inner.pyx":413 * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() # <<<<<<<<<<<<<< */ - __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_init); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 364, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_init); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 413, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_CallNoArg(__pyx_t_1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 364, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_CallNoArg(__pyx_t_1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 413, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_2) < 0) __PYX_ERR(0, 364, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_2) < 0) __PYX_ERR(0, 413, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; /* "average_inner.pyx":1 diff --git a/fse/models/average_inner.pxd b/fse/models/average_inner.pxd index ed245ae..2b17205 100644 --- a/fse/models/average_inner.pxd +++ b/fse/models/average_inner.pxd @@ -17,8 +17,21 @@ ctypedef np.float32_t REAL_t ctypedef np.uint32_t uINT_t # BLAS routine signatures -ctypedef void (*saxpy_ptr) (const int *N, const float *alpha, const float *X, const int *incX, float *Y, const int *incY) nogil -ctypedef void (*sscal_ptr) (const int *N, const float *alpha, const float *X, const int *incX) nogil +ctypedef void (*saxpy_ptr) ( + const int *N, + const float *alpha, + const float *X, + const int *incX, + float *Y, + const int *incY, +) nogil + +ctypedef void (*sscal_ptr) ( + const int *N, + const float *alpha, + const float *X, + const int *incX, +) nogil cdef saxpy_ptr saxpy cdef sscal_ptr sscal @@ -69,8 +82,26 @@ cdef struct FTSentenceVecsConfig: uINT_t subwords_idx_len[MAX_WORDS] uINT_t *subwords_idx -cdef init_base_s2v_config(BaseSentenceVecsConfig *c, model, target, memory) -cdef init_ft_s2v_config(FTSentenceVecsConfig *c, model, target, memory) - -cdef object populate_base_s2v_config(BaseSentenceVecsConfig *c, vocab, indexed_sentences) -cdef object populate_ft_s2v_config(FTSentenceVecsConfig *c, vocab, indexed_sentences) \ No newline at end of file +cdef init_base_s2v_config( + BaseSentenceVecsConfig *c, + model, + target, + memory +) +cdef init_ft_s2v_config( + FTSentenceVecsConfig *c, + model, + target, + memory +) + +cdef object populate_base_s2v_config( + BaseSentenceVecsConfig *c, + vocab, + indexed_sentences, +) +cdef object populate_ft_s2v_config( + FTSentenceVecsConfig *c, + vocab, + indexed_sentences, +) \ No newline at end of file diff --git a/fse/models/average_inner.pyx b/fse/models/average_inner.pyx index f301b93..b359eaa 100644 --- a/fse/models/average_inner.pyx +++ b/fse/models/average_inner.pyx @@ -34,7 +34,12 @@ cdef REAL_t ZEROF = 0.0 DEF MAX_WORDS = 10000 DEF MAX_NGRAMS = 40 -cdef init_base_s2v_config(BaseSentenceVecsConfig *c, model, target, memory): +cdef init_base_s2v_config( + BaseSentenceVecsConfig *c, + model, + target, + memory, +): """Load BaseAny2Vec parameters into a BaseSentenceVecsConfig struct. Parameters @@ -61,7 +66,12 @@ cdef init_base_s2v_config(BaseSentenceVecsConfig *c, model, target, memory): c[0].sentence_vectors = (np.PyArray_DATA(target)) -cdef init_ft_s2v_config(FTSentenceVecsConfig *c, model, target, memory): +cdef init_ft_s2v_config( + FTSentenceVecsConfig *c, + model, + target, + memory, +): """Load Fasttext parameters into a FTSentenceVecsConfig struct. Parameters @@ -98,7 +108,11 @@ cdef init_ft_s2v_config(FTSentenceVecsConfig *c, model, target, memory): c[0].sentence_vectors = (np.PyArray_DATA(target)) -cdef object populate_base_s2v_config(BaseSentenceVecsConfig *c, vocab, indexed_sentences): +cdef object populate_base_s2v_config( + BaseSentenceVecsConfig *c, + vocab, + indexed_sentences, +): """Prepare C structures for BaseAny2VecModel so we can go "full C" and release the Python GIL. We create indices over the sentences. We also perform some calculations for @@ -148,7 +162,11 @@ cdef object populate_base_s2v_config(BaseSentenceVecsConfig *c, vocab, indexed_s return eff_sents, eff_words -cdef object populate_ft_s2v_config(FTSentenceVecsConfig *c, vocab, indexed_sentences): +cdef object populate_ft_s2v_config( + FTSentenceVecsConfig *c, + vocab, + indexed_sentences, +): """Prepare C structures for FastText so we can go "full C" and release the Python GIL. We create indices over the sentences. We also perform some calculations for @@ -211,7 +229,10 @@ cdef object populate_ft_s2v_config(FTSentenceVecsConfig *c, vocab, indexed_sente return eff_sents, eff_words -cdef void compute_base_sentence_averages(BaseSentenceVecsConfig *c, uINT_t num_sentences) nogil: +cdef void compute_base_sentence_averages( + BaseSentenceVecsConfig *c, + uINT_t num_sentences, +) nogil: """Perform optimized sentence-level averaging for BaseAny2Vec model. Parameters @@ -256,7 +277,10 @@ cdef void compute_base_sentence_averages(BaseSentenceVecsConfig *c, uINT_t num_s # because it doesn't rescale the overall result saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) -cdef void compute_ft_sentence_averages(FTSentenceVecsConfig *c, uINT_t num_sentences) nogil: +cdef void compute_ft_sentence_averages( + FTSentenceVecsConfig *c, + uINT_t num_sentences, +) nogil: """Perform optimized sentence-level averaging for FastText model. Parameters @@ -311,7 +335,12 @@ cdef void compute_ft_sentence_averages(FTSentenceVecsConfig *c, uINT_t num_sente inv_count = ONEF / sent_len saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) -def train_average_cy(model, indexed_sentences, target, memory): +def train_average_cy( + model, + indexed_sentences, + target, + memory +): """Training on a sequence of sentences and update the target ndarray. Called internally from :meth:`~fse.models.average.Average._do_train_job`. @@ -343,17 +372,36 @@ def train_average_cy(model, indexed_sentences, target, memory): if not model.is_ft: init_base_s2v_config(&w2v, model, target, memory) - eff_sentences, eff_words = populate_base_s2v_config(&w2v, model.wv.vocab, indexed_sentences) + eff_sentences, eff_words = populate_base_s2v_config( + &w2v, + model.wv.vocab, + indexed_sentences, + ) with nogil: - compute_base_sentence_averages(&w2v, eff_sentences) + compute_base_sentence_averages( + &w2v, + eff_sentences, + ) else: - init_ft_s2v_config(&ft, model, target, memory) - - eff_sentences, eff_words = populate_ft_s2v_config(&ft, model.wv.vocab, indexed_sentences) + init_ft_s2v_config( + &ft, + model, + target, + memory, + ) + + eff_sentences, eff_words = populate_ft_s2v_config( + &ft, + model.wv.vocab, + indexed_sentences, + ) with nogil: - compute_ft_sentence_averages(&ft, eff_sentences) + compute_ft_sentence_averages( + &ft, + eff_sentences + ) return eff_sentences, eff_words diff --git a/fse/models/pooling_inner.pyx b/fse/models/pooling_inner.pyx index 0ba50cf..b04419a 100644 --- a/fse/models/pooling_inner.pyx +++ b/fse/models/pooling_inner.pyx @@ -219,7 +219,6 @@ cdef void compute_ft_sentence_pooling( ) # There's nothing to do here for many-to-one mappings - def train_pooling_cy( model, indexed_sentences, From eb0d190f952a385cbb94b03e5025b3165d3df0db Mon Sep 17 00:00:00 2001 From: OliverB Date: Fri, 28 Feb 2020 11:53:57 +0100 Subject: [PATCH 30/63] Fixed hpool base + added swrmax --- fse/models/average_inner.c | 234 +++++++------- fse/models/average_inner.pyx | 19 +- fse/models/pooling_inner.c | 585 +++++++++++++++++++++++------------ fse/models/pooling_inner.pyx | 86 ++++- fse/test/test_pooling.py | 35 ++- 5 files changed, 609 insertions(+), 350 deletions(-) diff --git a/fse/models/average_inner.c b/fse/models/average_inner.c index cb3ff81..79e6301 100644 --- a/fse/models/average_inner.c +++ b/fse/models/average_inner.c @@ -3591,22 +3591,22 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ * word_row = c.word_indices[i] * size * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< * - * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + * saxpy( */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); /* "average_inner.pyx":272 * word_idx = c.word_indices[i] * - * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) # <<<<<<<<<<<<<< - * - * if sent_len > ZEROF: + * saxpy( # <<<<<<<<<<<<<< + * &size, + * &c.word_weights[word_idx], */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); } - /* "average_inner.pyx":274 - * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + /* "average_inner.pyx":281 + * ) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< * inv_count = ONEF / sent_len @@ -3615,7 +3615,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_t_8 = ((__pyx_v_sent_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_8) { - /* "average_inner.pyx":275 + /* "average_inner.pyx":282 * * if sent_len > ZEROF: * inv_count = ONEF / sent_len # <<<<<<<<<<<<<< @@ -3624,17 +3624,17 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_sent_len); - /* "average_inner.pyx":278 + /* "average_inner.pyx":285 * # If we perform the a*x on memory, the computation is compatible with many-to-one mappings * # because it doesn't rescale the overall result - * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) # <<<<<<<<<<<<<< - * - * cdef void compute_ft_sentence_averages( + * saxpy( # <<<<<<<<<<<<<< + * &size, + * &inv_count, */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&__pyx_v_13average_inner_ONE)); - /* "average_inner.pyx":274 - * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + /* "average_inner.pyx":281 + * ) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< * inv_count = ONEF / sent_len @@ -3654,8 +3654,8 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ /* function exit code */ } -/* "average_inner.pyx":280 - * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) +/* "average_inner.pyx":294 + * ) * * cdef void compute_ft_sentence_averages( # <<<<<<<<<<<<<< * FTSentenceVecsConfig *c, @@ -3691,7 +3691,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_13average_inner_uINT_t __pyx_t_11; __pyx_t_13average_inner_uINT_t __pyx_t_12; - /* "average_inner.pyx":299 + /* "average_inner.pyx":313 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -3701,7 +3701,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "average_inner.pyx":309 + /* "average_inner.pyx":323 * REAL_t sent_len * REAL_t inv_count, inv_ngram * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< @@ -3711,7 +3711,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_2 = __pyx_v_c->oov_weight; __pyx_v_oov_weight = __pyx_t_2; - /* "average_inner.pyx":312 + /* "average_inner.pyx":326 * * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -3723,7 +3723,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { __pyx_v_sent_idx = __pyx_t_5; - /* "average_inner.pyx":313 + /* "average_inner.pyx":327 * * for sent_idx in range(num_sentences): * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -3732,7 +3732,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "average_inner.pyx":314 + /* "average_inner.pyx":328 * for sent_idx in range(num_sentences): * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -3741,7 +3741,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "average_inner.pyx":315 + /* "average_inner.pyx":329 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -3750,7 +3750,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "average_inner.pyx":316 + /* "average_inner.pyx":330 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -3759,7 +3759,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "average_inner.pyx":318 + /* "average_inner.pyx":332 * sent_len = ZEROF * * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -3771,7 +3771,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { __pyx_v_i = __pyx_t_8; - /* "average_inner.pyx":319 + /* "average_inner.pyx":333 * * for i in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -3780,7 +3780,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "average_inner.pyx":320 + /* "average_inner.pyx":334 * for i in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< @@ -3789,7 +3789,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); - /* "average_inner.pyx":322 + /* "average_inner.pyx":336 * sent_row = c.sent_adresses[i] * size * * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< @@ -3798,7 +3798,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); - /* "average_inner.pyx":323 + /* "average_inner.pyx":337 * * word_idx = c.word_indices[i] * ngrams = c.subwords_idx_len[i] # <<<<<<<<<<<<<< @@ -3807,7 +3807,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_i]); - /* "average_inner.pyx":325 + /* "average_inner.pyx":339 * ngrams = c.subwords_idx_len[i] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -3817,7 +3817,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); if (__pyx_t_9) { - /* "average_inner.pyx":326 + /* "average_inner.pyx":340 * * if ngrams == 0: * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< @@ -3826,7 +3826,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); - /* "average_inner.pyx":327 + /* "average_inner.pyx":341 * if ngrams == 0: * word_row = c.word_indices[i] * size * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) # <<<<<<<<<<<<<< @@ -3835,7 +3835,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); - /* "average_inner.pyx":325 + /* "average_inner.pyx":339 * ngrams = c.subwords_idx_len[i] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -3845,7 +3845,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ goto __pyx_L7; } - /* "average_inner.pyx":329 + /* "average_inner.pyx":343 * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) * else: * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< @@ -3855,7 +3855,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ /*else*/ { __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); - /* "average_inner.pyx":330 + /* "average_inner.pyx":344 * else: * inv_ngram = (ONEF / ngrams) * c.oov_weight * for j in range(ngrams): # <<<<<<<<<<<<<< @@ -3867,7 +3867,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ for (__pyx_t_12 = 0; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { __pyx_v_j = __pyx_t_12; - /* "average_inner.pyx":331 + /* "average_inner.pyx":345 * inv_ngram = (ONEF / ngrams) * c.oov_weight * for j in range(ngrams): * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size # <<<<<<<<<<<<<< @@ -3876,7 +3876,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_i * 40) + __pyx_v_j)]) * __pyx_v_size); - /* "average_inner.pyx":332 + /* "average_inner.pyx":346 * for j in range(ngrams): * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) # <<<<<<<<<<<<<< @@ -3889,7 +3889,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_L7:; } - /* "average_inner.pyx":334 + /* "average_inner.pyx":348 * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< @@ -3899,7 +3899,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_9 = ((__pyx_v_sent_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_9) { - /* "average_inner.pyx":335 + /* "average_inner.pyx":349 * * if sent_len > ZEROF: * inv_count = ONEF / sent_len # <<<<<<<<<<<<<< @@ -3908,7 +3908,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_sent_len); - /* "average_inner.pyx":336 + /* "average_inner.pyx":350 * if sent_len > ZEROF: * inv_count = ONEF / sent_len * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) # <<<<<<<<<<<<<< @@ -3917,7 +3917,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&__pyx_v_13average_inner_ONE)); - /* "average_inner.pyx":334 + /* "average_inner.pyx":348 * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< @@ -3927,8 +3927,8 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ } } - /* "average_inner.pyx":280 - * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + /* "average_inner.pyx":294 + * ) * * cdef void compute_ft_sentence_averages( # <<<<<<<<<<<<<< * FTSentenceVecsConfig *c, @@ -3938,7 +3938,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ /* function exit code */ } -/* "average_inner.pyx":338 +/* "average_inner.pyx":352 * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) * * def train_average_cy( # <<<<<<<<<<<<<< @@ -3985,23 +3985,23 @@ static PyObject *__pyx_pw_13average_inner_1train_average_cy(PyObject *__pyx_self case 1: if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 1); __PYX_ERR(0, 338, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 1); __PYX_ERR(0, 352, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 2: if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 2); __PYX_ERR(0, 338, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 2); __PYX_ERR(0, 352, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 3: if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 3); __PYX_ERR(0, 338, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 3); __PYX_ERR(0, 352, __pyx_L3_error) } } if (unlikely(kw_args > 0)) { - if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_average_cy") < 0)) __PYX_ERR(0, 338, __pyx_L3_error) + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_average_cy") < 0)) __PYX_ERR(0, 352, __pyx_L3_error) } } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { goto __pyx_L5_argtuple_error; @@ -4018,7 +4018,7 @@ static PyObject *__pyx_pw_13average_inner_1train_average_cy(PyObject *__pyx_self } goto __pyx_L4_argument_unpacking_done; __pyx_L5_argtuple_error:; - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 338, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 352, __pyx_L3_error) __pyx_L3_error:; __Pyx_AddTraceback("average_inner.train_average_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __Pyx_RefNannyFinishContext(); @@ -4049,7 +4049,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_t_13average_inner_uINT_t __pyx_t_9; __Pyx_RefNannySetupContext("train_average_cy", 0); - /* "average_inner.pyx":367 + /* "average_inner.pyx":381 * """ * * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< @@ -4058,7 +4058,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_sentences = 0; - /* "average_inner.pyx":368 + /* "average_inner.pyx":382 * * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< @@ -4067,52 +4067,52 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_words = 0; - /* "average_inner.pyx":372 + /* "average_inner.pyx":386 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< * init_base_s2v_config(&w2v, model, target, memory) * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 372, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 386, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 372, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 386, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_3 = ((!__pyx_t_2) != 0); if (__pyx_t_3) { - /* "average_inner.pyx":373 + /* "average_inner.pyx":387 * * if not model.is_ft: * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< * * eff_sentences, eff_words = populate_base_s2v_config( */ - __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 373, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 387, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":377 + /* "average_inner.pyx":391 * eff_sentences, eff_words = populate_base_s2v_config( * &w2v, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences, * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 377, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 391, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 377, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 391, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":375 + /* "average_inner.pyx":389 * init_base_s2v_config(&w2v, model, target, memory) * * eff_sentences, eff_words = populate_base_s2v_config( # <<<<<<<<<<<<<< * &w2v, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 375, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 389, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -4121,7 +4121,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 375, __pyx_L1_error) + __PYX_ERR(0, 389, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -4134,15 +4134,15 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_4); __Pyx_INCREF(__pyx_t_5); #else - __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 375, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 389, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 375, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 389, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 375, __pyx_L1_error) + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 389, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; @@ -4150,7 +4150,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_4); index = 1; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L4_unpacking_failed; __Pyx_GOTREF(__pyx_t_5); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 375, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 389, __pyx_L1_error) __pyx_t_7 = NULL; __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; goto __pyx_L5_unpacking_done; @@ -4158,17 +4158,17 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_t_7 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 375, __pyx_L1_error) + __PYX_ERR(0, 389, __pyx_L1_error) __pyx_L5_unpacking_done:; } - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 375, __pyx_L1_error) + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 389, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 375, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 389, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; __pyx_v_eff_sentences = __pyx_t_8; __pyx_v_eff_words = __pyx_t_9; - /* "average_inner.pyx":381 + /* "average_inner.pyx":395 * ) * * with nogil: # <<<<<<<<<<<<<< @@ -4183,7 +4183,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "average_inner.pyx":382 + /* "average_inner.pyx":396 * * with nogil: * compute_base_sentence_averages( # <<<<<<<<<<<<<< @@ -4193,7 +4193,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_f_13average_inner_compute_base_sentence_averages((&__pyx_v_w2v), __pyx_v_eff_sentences); } - /* "average_inner.pyx":381 + /* "average_inner.pyx":395 * ) * * with nogil: # <<<<<<<<<<<<<< @@ -4212,7 +4212,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec } } - /* "average_inner.pyx":372 + /* "average_inner.pyx":386 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< @@ -4222,7 +4222,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec goto __pyx_L3; } - /* "average_inner.pyx":387 + /* "average_inner.pyx":401 * ) * else: * init_ft_s2v_config( # <<<<<<<<<<<<<< @@ -4231,38 +4231,38 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec */ /*else*/ { - /* "average_inner.pyx":391 + /* "average_inner.pyx":405 * model, * target, * memory, # <<<<<<<<<<<<<< * ) * */ - __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 387, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 401, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":396 + /* "average_inner.pyx":410 * eff_sentences, eff_words = populate_ft_s2v_config( * &ft, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences, * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 396, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 410, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 396, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 410, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":394 + /* "average_inner.pyx":408 * ) * * eff_sentences, eff_words = populate_ft_s2v_config( # <<<<<<<<<<<<<< * &ft, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 394, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 408, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -4271,7 +4271,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 394, __pyx_L1_error) + __PYX_ERR(0, 408, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -4284,15 +4284,15 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_5); __Pyx_INCREF(__pyx_t_4); #else - __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 394, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 408, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); - __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 394, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 408, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 394, __pyx_L1_error) + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 408, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; @@ -4300,7 +4300,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_5); index = 1; __pyx_t_4 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_4)) goto __pyx_L9_unpacking_failed; __Pyx_GOTREF(__pyx_t_4); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 394, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 408, __pyx_L1_error) __pyx_t_7 = NULL; __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; goto __pyx_L10_unpacking_done; @@ -4308,17 +4308,17 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_t_7 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 394, __pyx_L1_error) + __PYX_ERR(0, 408, __pyx_L1_error) __pyx_L10_unpacking_done:; } - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 394, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 408, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 394, __pyx_L1_error) + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 408, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; __pyx_v_eff_sentences = __pyx_t_9; __pyx_v_eff_words = __pyx_t_8; - /* "average_inner.pyx":400 + /* "average_inner.pyx":414 * ) * * with nogil: # <<<<<<<<<<<<<< @@ -4333,7 +4333,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "average_inner.pyx":401 + /* "average_inner.pyx":415 * * with nogil: * compute_ft_sentence_averages( # <<<<<<<<<<<<<< @@ -4343,7 +4343,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_f_13average_inner_compute_ft_sentence_averages((&__pyx_v_ft), __pyx_v_eff_sentences); } - /* "average_inner.pyx":400 + /* "average_inner.pyx":414 * ) * * with nogil: # <<<<<<<<<<<<<< @@ -4364,7 +4364,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec } __pyx_L3:; - /* "average_inner.pyx":406 + /* "average_inner.pyx":420 * ) * * return eff_sentences, eff_words # <<<<<<<<<<<<<< @@ -4372,11 +4372,11 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec * def init(): */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 406, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 420, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 406, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 420, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 406, __pyx_L1_error) + __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 420, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_1); @@ -4388,7 +4388,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_t_5 = 0; goto __pyx_L0; - /* "average_inner.pyx":338 + /* "average_inner.pyx":352 * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) * * def train_average_cy( # <<<<<<<<<<<<<< @@ -4410,7 +4410,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec return __pyx_r; } -/* "average_inner.pyx":408 +/* "average_inner.pyx":422 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -4438,7 +4438,7 @@ static PyObject *__pyx_pf_13average_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("init", 0); - /* "average_inner.pyx":409 + /* "average_inner.pyx":423 * * def init(): * return 1 # <<<<<<<<<<<<<< @@ -4450,7 +4450,7 @@ static PyObject *__pyx_pf_13average_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __pyx_r = __pyx_int_1; goto __pyx_L0; - /* "average_inner.pyx":408 + /* "average_inner.pyx":422 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -7088,26 +7088,26 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { __Pyx_GOTREF(__pyx_tuple__7); __Pyx_GIVEREF(__pyx_tuple__7); - /* "average_inner.pyx":338 + /* "average_inner.pyx":352 * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) * * def train_average_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 338, __pyx_L1_error) + __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 352, __pyx_L1_error) __Pyx_GOTREF(__pyx_tuple__9); __Pyx_GIVEREF(__pyx_tuple__9); - __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_train_average_cy, 338, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 338, __pyx_L1_error) + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_train_average_cy, 352, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 352, __pyx_L1_error) - /* "average_inner.pyx":408 + /* "average_inner.pyx":422 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_init, 408, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 408, __pyx_L1_error) + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_init, 422, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 422, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -7563,58 +7563,58 @@ if (!__Pyx_RefNanny) { */ __pyx_v_13average_inner_ZEROF = ((__pyx_t_13average_inner_REAL_t)0.0); - /* "average_inner.pyx":338 + /* "average_inner.pyx":352 * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) * * def train_average_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_1train_average_cy, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 338, __pyx_L1_error) + __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_1train_average_cy, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 352, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_average_cy, __pyx_t_1) < 0) __PYX_ERR(0, 338, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_average_cy, __pyx_t_1) < 0) __PYX_ERR(0, 352, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":408 + /* "average_inner.pyx":422 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_3init, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 408, __pyx_L1_error) + __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_3init, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 422, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_1) < 0) __PYX_ERR(0, 408, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_1) < 0) __PYX_ERR(0, 422, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":411 + /* "average_inner.pyx":425 * return 1 * * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 411, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 425, __pyx_L1_error) - /* "average_inner.pyx":412 + /* "average_inner.pyx":426 * * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 412, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 426, __pyx_L1_error) - /* "average_inner.pyx":413 + /* "average_inner.pyx":427 * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() # <<<<<<<<<<<<<< */ - __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_init); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 413, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_init); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 427, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_CallNoArg(__pyx_t_1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 413, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_CallNoArg(__pyx_t_1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 427, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_2) < 0) __PYX_ERR(0, 413, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_2) < 0) __PYX_ERR(0, 427, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; /* "average_inner.pyx":1 diff --git a/fse/models/average_inner.pyx b/fse/models/average_inner.pyx index b359eaa..5ceb5e0 100644 --- a/fse/models/average_inner.pyx +++ b/fse/models/average_inner.pyx @@ -19,6 +19,7 @@ cimport numpy as np from gensim.models._utils_any2vec import compute_ngrams_bytes, ft_hash_bytes from libc.string cimport memset +from libc.stdio cimport printf import scipy.linalg.blas as fblas @@ -269,13 +270,27 @@ cdef void compute_base_sentence_averages( word_row = c.word_indices[i] * size word_idx = c.word_indices[i] - saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + saxpy( + &size, + &c.word_weights[word_idx], + &c.word_vectors[word_row], + &ONE, + c.mem, + &ONE + ) if sent_len > ZEROF: inv_count = ONEF / sent_len # If we perform the a*x on memory, the computation is compatible with many-to-one mappings # because it doesn't rescale the overall result - saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + saxpy( + &size, + &inv_count, + c.mem, + &ONE, + &c.sentence_vectors[sent_row], + &ONE + ) cdef void compute_ft_sentence_averages( FTSentenceVecsConfig *c, diff --git a/fse/models/pooling_inner.c b/fse/models/pooling_inner.c index 8164dc1..a5c26a2 100644 --- a/fse/models/pooling_inner.c +++ b/fse/models/pooling_inner.c @@ -1141,22 +1141,22 @@ struct __pyx_t_13average_inner_FTSentenceVecsConfig; /* "average_inner.pxd":20 * * # BLAS routine signatures - * ctypedef void (*saxpy_ptr) (const int *N, const float *alpha, const float *X, const int *incX, float *Y, const int *incY) nogil # <<<<<<<<<<<<<< - * ctypedef void (*sscal_ptr) (const int *N, const float *alpha, const float *X, const int *incX) nogil - * + * ctypedef void (*saxpy_ptr) ( # <<<<<<<<<<<<<< + * const int *N, + * const float *alpha, */ typedef void (*__pyx_t_13average_inner_saxpy_ptr)(int const *, float const *, float const *, int const *, float *, int const *); -/* "average_inner.pxd":21 - * # BLAS routine signatures - * ctypedef void (*saxpy_ptr) (const int *N, const float *alpha, const float *X, const int *incX, float *Y, const int *incY) nogil - * ctypedef void (*sscal_ptr) (const int *N, const float *alpha, const float *X, const int *incX) nogil # <<<<<<<<<<<<<< +/* "average_inner.pxd":29 + * ) nogil * - * cdef saxpy_ptr saxpy + * ctypedef void (*sscal_ptr) ( # <<<<<<<<<<<<<< + * const int *N, + * const float *alpha, */ typedef void (*__pyx_t_13average_inner_sscal_ptr)(int const *, float const *, float const *, int const *); -/* "average_inner.pxd":35 +/* "average_inner.pxd":48 * DEF MAX_NGRAMS = 40 * * cdef struct BaseSentenceVecsConfig: # <<<<<<<<<<<<<< @@ -1176,7 +1176,7 @@ struct __pyx_t_13average_inner_BaseSentenceVecsConfig { __pyx_t_13average_inner_uINT_t sentence_boundary[(0x2710 + 1)]; }; -/* "average_inner.pxd":49 +/* "average_inner.pxd":62 * uINT_t sentence_boundary[MAX_WORDS + 1] * * cdef struct FTSentenceVecsConfig: # <<<<<<<<<<<<<< @@ -1723,7 +1723,7 @@ static PyObject *(*__pyx_f_13average_inner_populate_base_s2v_config)(struct __py static PyObject *(*__pyx_f_13average_inner_populate_ft_s2v_config)(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *); /*proto*/ /* Module declarations from 'pooling_inner' */ -static void __pyx_f_13pooling_inner_sl_max_pool(int const *, float *, float const *); /*proto*/ +static void __pyx_f_13pooling_inner_swrmax_pool(int const *, float const *, float const *, float *); /*proto*/ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_uINT_t); /*proto*/ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ @@ -1840,71 +1840,71 @@ static PyObject *__pyx_codeobj__10; static PyObject *__pyx_codeobj__11; /* Late includes */ -/* "pooling_inner.pyx":43 +/* "pooling_inner.pyx":44 * DEF MAX_NGRAMS = 40 * - * cdef void sl_max_pool( # <<<<<<<<<<<<<< + * cdef void swrmax_pool( # <<<<<<<<<<<<<< * const int *N, - * float *X, + * const float *alpha, */ -static void __pyx_f_13pooling_inner_sl_max_pool(int const *__pyx_v_N, float *__pyx_v_X, float const *__pyx_v_Y) { +static void __pyx_f_13pooling_inner_swrmax_pool(int const *__pyx_v_N, float const *__pyx_v_alpha, float const *__pyx_v_X, float *__pyx_v_Y) { int __pyx_v_i; int __pyx_t_1; int __pyx_t_2; - /* "pooling_inner.pyx":61 + /* "pooling_inner.pyx":65 * """ * cdef int i * for i from 0 <= i < N[0] by 1: # <<<<<<<<<<<<<< - * if X[i] < Y[i]: - * X[i] = Y[i] + * if (alpha[0] * X[i]) > Y[i]: + * Y[i] = alpha[0] * X[i] */ __pyx_t_1 = (__pyx_v_N[0]); for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_1; __pyx_v_i+=1) { - /* "pooling_inner.pyx":62 + /* "pooling_inner.pyx":66 * cdef int i * for i from 0 <= i < N[0] by 1: - * if X[i] < Y[i]: # <<<<<<<<<<<<<< - * X[i] = Y[i] + * if (alpha[0] * X[i]) > Y[i]: # <<<<<<<<<<<<<< + * Y[i] = alpha[0] * X[i] * */ - __pyx_t_2 = (((__pyx_v_X[__pyx_v_i]) < (__pyx_v_Y[__pyx_v_i])) != 0); + __pyx_t_2 = ((((__pyx_v_alpha[0]) * (__pyx_v_X[__pyx_v_i])) > (__pyx_v_Y[__pyx_v_i])) != 0); if (__pyx_t_2) { - /* "pooling_inner.pyx":63 + /* "pooling_inner.pyx":67 * for i from 0 <= i < N[0] by 1: - * if X[i] < Y[i]: - * X[i] = Y[i] # <<<<<<<<<<<<<< + * if (alpha[0] * X[i]) > Y[i]: + * Y[i] = alpha[0] * X[i] # <<<<<<<<<<<<<< * * cdef void compute_base_sentence_pooling( */ - (__pyx_v_X[__pyx_v_i]) = (__pyx_v_Y[__pyx_v_i]); + (__pyx_v_Y[__pyx_v_i]) = ((__pyx_v_alpha[0]) * (__pyx_v_X[__pyx_v_i])); - /* "pooling_inner.pyx":62 + /* "pooling_inner.pyx":66 * cdef int i * for i from 0 <= i < N[0] by 1: - * if X[i] < Y[i]: # <<<<<<<<<<<<<< - * X[i] = Y[i] + * if (alpha[0] * X[i]) > Y[i]: # <<<<<<<<<<<<<< + * Y[i] = alpha[0] * X[i] * */ } } - /* "pooling_inner.pyx":43 + /* "pooling_inner.pyx":44 * DEF MAX_NGRAMS = 40 * - * cdef void sl_max_pool( # <<<<<<<<<<<<<< + * cdef void swrmax_pool( # <<<<<<<<<<<<<< * const int *N, - * float *X, + * const float *alpha, */ /* function exit code */ } -/* "pooling_inner.pyx":65 - * X[i] = Y[i] +/* "pooling_inner.pyx":69 + * Y[i] = alpha[0] * X[i] * * cdef void compute_base_sentence_pooling( # <<<<<<<<<<<<<< * BaseSentenceVecsConfig *c, @@ -1918,6 +1918,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t __pyx_t_13average_inner_uINT_t __pyx_v_sent_end; __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; __pyx_t_13average_inner_uINT_t __pyx_v_i; + __pyx_t_13average_inner_uINT_t __pyx_v_word_idx; __pyx_t_13average_inner_uINT_t __pyx_v_word_row; __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; int __pyx_t_1; @@ -1928,7 +1929,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t __pyx_t_13average_inner_uINT_t __pyx_t_6; __pyx_t_13average_inner_uINT_t __pyx_t_7; - /* "pooling_inner.pyx":84 + /* "pooling_inner.pyx":88 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -1938,7 +1939,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":92 + /* "pooling_inner.pyx":96 * REAL_t sent_len, inv_count * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -1950,7 +1951,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) { __pyx_v_sent_idx = __pyx_t_4; - /* "pooling_inner.pyx":93 + /* "pooling_inner.pyx":97 * * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -1959,7 +1960,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":94 + /* "pooling_inner.pyx":98 * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -1968,7 +1969,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":95 + /* "pooling_inner.pyx":99 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -1977,7 +1978,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":97 + /* "pooling_inner.pyx":101 * sent_len = ZEROF * * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -1989,7 +1990,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { __pyx_v_i = __pyx_t_7; - /* "pooling_inner.pyx":98 + /* "pooling_inner.pyx":102 * * for i in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -1998,37 +1999,46 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":99 + /* "pooling_inner.pyx":103 * for i in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< * word_row = c.word_indices[i] * size - * + * word_idx = c.word_indices[i] */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); - /* "pooling_inner.pyx":100 + /* "pooling_inner.pyx":104 * sent_len += ONEF * sent_row = c.sent_adresses[i] * size * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< + * word_idx = c.word_indices[i] * - * sl_max_pool( */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); - /* "pooling_inner.pyx":102 + /* "pooling_inner.pyx":105 + * sent_row = c.sent_adresses[i] * size * word_row = c.word_indices[i] * size + * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< * - * sl_max_pool( # <<<<<<<<<<<<<< + * swrmax_pool( + */ + __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); + + /* "pooling_inner.pyx":107 + * word_idx = c.word_indices[i] + * + * swrmax_pool( # <<<<<<<<<<<<<< * &size, - * &c.sentence_vectors[sent_row], + * &c.word_weights[word_idx], */ - __pyx_f_13pooling_inner_sl_max_pool((&__pyx_v_size), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row]))); + __pyx_f_13pooling_inner_swrmax_pool((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row]))); } } - /* "pooling_inner.pyx":65 - * X[i] = Y[i] + /* "pooling_inner.pyx":69 + * Y[i] = alpha[0] * X[i] * * cdef void compute_base_sentence_pooling( # <<<<<<<<<<<<<< * BaseSentenceVecsConfig *c, @@ -2038,7 +2048,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t /* function exit code */ } -/* "pooling_inner.pyx":110 +/* "pooling_inner.pyx":116 * * * cdef void compute_base_sentence_hier_pooling( # <<<<<<<<<<<<<< @@ -2046,15 +2056,20 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t * uINT_t num_sentences, */ -static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences, CYTHON_UNUSED __pyx_t_13average_inner_uINT_t __pyx_v_window_size) { +static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences, __pyx_t_13average_inner_uINT_t __pyx_v_window_size) { int __pyx_v_size; __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; __pyx_t_13average_inner_uINT_t __pyx_v_sent_end; - CYTHON_UNUSED __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; + __pyx_t_13average_inner_uINT_t __pyx_v_window_end; __pyx_t_13average_inner_uINT_t __pyx_v_i; - CYTHON_UNUSED __pyx_t_13average_inner_uINT_t __pyx_v_word_row; + __pyx_t_13average_inner_uINT_t __pyx_v_j; + __pyx_t_13average_inner_uINT_t __pyx_v_word_idx; + __pyx_t_13average_inner_uINT_t __pyx_v_word_row; __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; + __pyx_t_13average_inner_REAL_t __pyx_v_win_len; + __pyx_t_13average_inner_REAL_t __pyx_v_inv_count; int __pyx_t_1; __pyx_t_13average_inner_uINT_t __pyx_t_2; __pyx_t_13average_inner_uINT_t __pyx_t_3; @@ -2062,19 +2077,23 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ __pyx_t_13average_inner_uINT_t __pyx_t_5; __pyx_t_13average_inner_uINT_t __pyx_t_6; __pyx_t_13average_inner_uINT_t __pyx_t_7; + int __pyx_t_8; + __pyx_t_13average_inner_uINT_t __pyx_t_9; + __pyx_t_13average_inner_uINT_t __pyx_t_10; + __pyx_t_13average_inner_uINT_t __pyx_t_11; - /* "pooling_inner.pyx":130 - * """ + /* "pooling_inner.pyx":139 + * * cdef: * int size = c.size # <<<<<<<<<<<<<< * - * uINT_t sent_idx, sent_start, sent_end, sent_row + * uINT_t sent_idx, sent_start, sent_end, sent_row, window_end */ __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":138 - * REAL_t sent_len, inv_count + /* "pooling_inner.pyx":147 + * REAL_t sent_len, win_len, inv_count * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< * sent_start = c.sentence_boundary[sent_idx] @@ -2085,7 +2104,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) { __pyx_v_sent_idx = __pyx_t_4; - /* "pooling_inner.pyx":139 + /* "pooling_inner.pyx":148 * * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -2094,7 +2113,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":140 + /* "pooling_inner.pyx":149 * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -2103,7 +2122,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":141 + /* "pooling_inner.pyx":150 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -2112,7 +2131,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":143 + /* "pooling_inner.pyx":152 * sent_len = ZEROF * * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -2124,36 +2143,190 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { __pyx_v_i = __pyx_t_7; - /* "pooling_inner.pyx":144 + /* "pooling_inner.pyx":153 * * for i in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< * sent_row = c.sent_adresses[i] * size - * word_row = c.word_indices[i] * size + * */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":145 + /* "pooling_inner.pyx":154 * for i in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< - * word_row = c.word_indices[i] * size * + * if i + window_size > sent_end: */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); - /* "pooling_inner.pyx":146 - * sent_len += ONEF + /* "pooling_inner.pyx":156 * sent_row = c.sent_adresses[i] * size - * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< * + * if i + window_size > sent_end: # <<<<<<<<<<<<<< + * window_end = sent_end + * else: + */ + __pyx_t_8 = (((__pyx_v_i + __pyx_v_window_size) > __pyx_v_sent_end) != 0); + if (__pyx_t_8) { + + /* "pooling_inner.pyx":157 * + * if i + window_size > sent_end: + * window_end = sent_end # <<<<<<<<<<<<<< + * else: + * window_end = i + window_size */ - __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); + __pyx_v_window_end = __pyx_v_sent_end; + + /* "pooling_inner.pyx":156 + * sent_row = c.sent_adresses[i] * size + * + * if i + window_size > sent_end: # <<<<<<<<<<<<<< + * window_end = sent_end + * else: + */ + goto __pyx_L7; + } + + /* "pooling_inner.pyx":159 + * window_end = sent_end + * else: + * window_end = i + window_size # <<<<<<<<<<<<<< + * + * # Compute the locally averaged window + */ + /*else*/ { + __pyx_v_window_end = (__pyx_v_i + __pyx_v_window_size); + } + __pyx_L7:; + + /* "pooling_inner.pyx":162 + * + * # Compute the locally averaged window + * win_len = ZEROF # <<<<<<<<<<<<<< + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + */ + __pyx_v_win_len = __pyx_v_13average_inner_ZEROF; + + /* "pooling_inner.pyx":163 + * # Compute the locally averaged window + * win_len = ZEROF + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + * for j in range(i, window_end): + */ + (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); + + /* "pooling_inner.pyx":164 + * win_len = ZEROF + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< + * for j in range(i, window_end): + * win_len += ONEF + */ + (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); + + /* "pooling_inner.pyx":165 + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + * for j in range(i, window_end): # <<<<<<<<<<<<<< + * win_len += ONEF + * + */ + __pyx_t_9 = __pyx_v_window_end; + __pyx_t_10 = __pyx_t_9; + for (__pyx_t_11 = __pyx_v_i; __pyx_t_11 < __pyx_t_10; __pyx_t_11+=1) { + __pyx_v_j = __pyx_t_11; + + /* "pooling_inner.pyx":166 + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + * for j in range(i, window_end): + * win_len += ONEF # <<<<<<<<<<<<<< + * + * word_row = c.word_indices[j] * size + */ + __pyx_v_win_len = (__pyx_v_win_len + __pyx_v_13average_inner_ONEF); + + /* "pooling_inner.pyx":168 + * win_len += ONEF + * + * word_row = c.word_indices[j] * size # <<<<<<<<<<<<<< + * word_idx = c.word_indices[j] + * + */ + __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_j]) * __pyx_v_size); + + /* "pooling_inner.pyx":169 + * + * word_row = c.word_indices[j] * size + * word_idx = c.word_indices[j] # <<<<<<<<<<<<<< + * + * saxpy( + */ + __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_j]); + + /* "pooling_inner.pyx":171 + * word_idx = c.word_indices[j] + * + * saxpy( # <<<<<<<<<<<<<< + * &size, + * &c.word_weights[word_idx], + */ + __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); + } + + /* "pooling_inner.pyx":181 + * + * # Rescale for dynamic window size + * if win_len > ZEROF: # <<<<<<<<<<<<<< + * inv_count = ONEF / win_len + * saxpy( + */ + __pyx_t_8 = ((__pyx_v_win_len > __pyx_v_13average_inner_ZEROF) != 0); + if (__pyx_t_8) { + + /* "pooling_inner.pyx":182 + * # Rescale for dynamic window size + * if win_len > ZEROF: + * inv_count = ONEF / win_len # <<<<<<<<<<<<<< + * saxpy( + * &size, + */ + __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_win_len); + + /* "pooling_inner.pyx":183 + * if win_len > ZEROF: + * inv_count = ONEF / win_len + * saxpy( # <<<<<<<<<<<<<< + * &size, + * &inv_count, + */ + __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE)); + + /* "pooling_inner.pyx":181 + * + * # Rescale for dynamic window size + * if win_len > ZEROF: # <<<<<<<<<<<<<< + * inv_count = ONEF / win_len + * saxpy( + */ + } + + /* "pooling_inner.pyx":192 + * ) + * + * swrmax_pool( # <<<<<<<<<<<<<< + * &size, + * &ONEF, + */ + __pyx_f_13pooling_inner_swrmax_pool((&__pyx_v_size), (&__pyx_v_13average_inner_ONEF), __pyx_v_c->mem2, (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row]))); } } - /* "pooling_inner.pyx":110 + /* "pooling_inner.pyx":116 * * * cdef void compute_base_sentence_hier_pooling( # <<<<<<<<<<<<<< @@ -2164,8 +2337,8 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ /* function exit code */ } -/* "pooling_inner.pyx":149 - * +/* "pooling_inner.pyx":199 + * ) * * cdef void compute_ft_sentence_pooling( # <<<<<<<<<<<<<< * FTSentenceVecsConfig *c, @@ -2182,11 +2355,11 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_13average_inner_uINT_t __pyx_v_ngrams; __pyx_t_13average_inner_uINT_t __pyx_v_i; __pyx_t_13average_inner_uINT_t __pyx_v_j; - CYTHON_UNUSED __pyx_t_13average_inner_uINT_t __pyx_v_word_idx; + __pyx_t_13average_inner_uINT_t __pyx_v_word_idx; __pyx_t_13average_inner_uINT_t __pyx_v_word_row; __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; __pyx_t_13average_inner_REAL_t __pyx_v_inv_ngram; - CYTHON_UNUSED __pyx_t_13average_inner_REAL_t __pyx_v_oov_weight; + __pyx_t_13average_inner_REAL_t __pyx_v_oov_weight; int __pyx_t_1; __pyx_t_13average_inner_REAL_t __pyx_t_2; __pyx_t_13average_inner_uINT_t __pyx_t_3; @@ -2200,7 +2373,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_13average_inner_uINT_t __pyx_t_11; __pyx_t_13average_inner_uINT_t __pyx_t_12; - /* "pooling_inner.pyx":168 + /* "pooling_inner.pyx":218 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -2210,7 +2383,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":178 + /* "pooling_inner.pyx":228 * REAL_t sent_len * REAL_t inv_count, inv_ngram * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< @@ -2220,7 +2393,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_2 = __pyx_v_c->oov_weight; __pyx_v_oov_weight = __pyx_t_2; - /* "pooling_inner.pyx":180 + /* "pooling_inner.pyx":230 * REAL_t oov_weight = c.oov_weight * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -2232,7 +2405,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { __pyx_v_sent_idx = __pyx_t_5; - /* "pooling_inner.pyx":181 + /* "pooling_inner.pyx":231 * * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -2241,7 +2414,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":182 + /* "pooling_inner.pyx":232 * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -2250,7 +2423,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":183 + /* "pooling_inner.pyx":233 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -2259,7 +2432,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":185 + /* "pooling_inner.pyx":235 * sent_len = ZEROF * * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -2271,7 +2444,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { __pyx_v_i = __pyx_t_8; - /* "pooling_inner.pyx":186 + /* "pooling_inner.pyx":236 * * for i in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -2280,7 +2453,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":187 + /* "pooling_inner.pyx":237 * for i in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< @@ -2289,7 +2462,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); - /* "pooling_inner.pyx":189 + /* "pooling_inner.pyx":239 * sent_row = c.sent_adresses[i] * size * * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< @@ -2298,7 +2471,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); - /* "pooling_inner.pyx":190 + /* "pooling_inner.pyx":240 * * word_idx = c.word_indices[i] * ngrams = c.subwords_idx_len[i] # <<<<<<<<<<<<<< @@ -2307,7 +2480,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_i]); - /* "pooling_inner.pyx":192 + /* "pooling_inner.pyx":242 * ngrams = c.subwords_idx_len[i] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2317,25 +2490,25 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":193 + /* "pooling_inner.pyx":243 * * if ngrams == 0: * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< * - * sl_max_pool( + * swrmax_pool( */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); - /* "pooling_inner.pyx":195 + /* "pooling_inner.pyx":245 * word_row = c.word_indices[i] * size * - * sl_max_pool( # <<<<<<<<<<<<<< + * swrmax_pool( # <<<<<<<<<<<<<< * &size, - * &c.sentence_vectors[sent_row], + * &c.word_weights[word_idx], */ - __pyx_f_13pooling_inner_sl_max_pool((&__pyx_v_size), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row]))); + __pyx_f_13pooling_inner_swrmax_pool((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row]))); - /* "pooling_inner.pyx":192 + /* "pooling_inner.pyx":242 * ngrams = c.subwords_idx_len[i] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2345,7 +2518,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 goto __pyx_L7; } - /* "pooling_inner.pyx":202 + /* "pooling_inner.pyx":253 * * else: * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2355,7 +2528,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 /*else*/ { (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":203 + /* "pooling_inner.pyx":254 * else: * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< @@ -2364,7 +2537,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); - /* "pooling_inner.pyx":204 + /* "pooling_inner.pyx":255 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight * for j in range(ngrams): # <<<<<<<<<<<<<< @@ -2376,7 +2549,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 for (__pyx_t_12 = 0; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { __pyx_v_j = __pyx_t_12; - /* "pooling_inner.pyx":205 + /* "pooling_inner.pyx":256 * inv_ngram = (ONEF / ngrams) * c.oov_weight * for j in range(ngrams): * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size # <<<<<<<<<<<<<< @@ -2385,7 +2558,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_i * 40) + __pyx_v_j)]) * __pyx_v_size); - /* "pooling_inner.pyx":206 + /* "pooling_inner.pyx":257 * for j in range(ngrams): * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size * saxpy( # <<<<<<<<<<<<<< @@ -2395,21 +2568,21 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_ngram), (&(__pyx_v_c->ngram_vectors[__pyx_v_ngram_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); } - /* "pooling_inner.pyx":215 + /* "pooling_inner.pyx":266 * ) * - * sl_max_pool( # <<<<<<<<<<<<<< + * swrmax_pool( # <<<<<<<<<<<<<< * &size, - * &c.sentence_vectors[sent_row], + * &oov_weight, */ - __pyx_f_13pooling_inner_sl_max_pool((&__pyx_v_size), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), __pyx_v_c->mem); + __pyx_f_13pooling_inner_swrmax_pool((&__pyx_v_size), (&__pyx_v_oov_weight), __pyx_v_c->mem, (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row]))); } __pyx_L7:; } } - /* "pooling_inner.pyx":149 - * + /* "pooling_inner.pyx":199 + * ) * * cdef void compute_ft_sentence_pooling( # <<<<<<<<<<<<<< * FTSentenceVecsConfig *c, @@ -2419,8 +2592,8 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 /* function exit code */ } -/* "pooling_inner.pyx":223 - * +/* "pooling_inner.pyx":274 + * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, @@ -2466,23 +2639,23 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self case 1: if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 223, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 274, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 2: if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 223, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 274, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 3: if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 223, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 274, __pyx_L3_error) } } if (unlikely(kw_args > 0)) { - if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 223, __pyx_L3_error) + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 274, __pyx_L3_error) } } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { goto __pyx_L5_argtuple_error; @@ -2499,7 +2672,7 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self } goto __pyx_L4_argument_unpacking_done; __pyx_L5_argtuple_error:; - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 223, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 274, __pyx_L3_error) __pyx_L3_error:; __Pyx_AddTraceback("pooling_inner.train_pooling_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __Pyx_RefNannyFinishContext(); @@ -2531,7 +2704,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_13average_inner_uINT_t __pyx_t_9; __Pyx_RefNannySetupContext("train_pooling_cy", 0); - /* "pooling_inner.pyx":252 + /* "pooling_inner.pyx":303 * """ * * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< @@ -2540,7 +2713,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_sentences = 0; - /* "pooling_inner.pyx":253 + /* "pooling_inner.pyx":304 * * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< @@ -2549,65 +2722,65 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_words = 0; - /* "pooling_inner.pyx":254 + /* "pooling_inner.pyx":305 * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 * cdef uINT_t window_size = model.window_size # <<<<<<<<<<<<<< * cdef BaseSentenceVecsConfig w2v * cdef FTSentenceVecsConfig ft */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_size); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 254, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_size); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 305, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_1); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 254, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_1); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 305, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_v_window_size = ((__pyx_t_13average_inner_uINT_t)__pyx_t_2); - /* "pooling_inner.pyx":258 + /* "pooling_inner.pyx":309 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< * init_base_s2v_config(&w2v, model, target, memory) * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 258, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 309, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 258, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 309, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_4 = ((!__pyx_t_3) != 0); if (__pyx_t_4) { - /* "pooling_inner.pyx":259 + /* "pooling_inner.pyx":310 * * if not model.is_ft: * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< * * eff_sentences, eff_words = populate_base_s2v_config( */ - __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 259, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 310, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":263 + /* "pooling_inner.pyx":314 * eff_sentences, eff_words = populate_base_s2v_config( * &w2v, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 263, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 314, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 263, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 314, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":261 + /* "pooling_inner.pyx":312 * init_base_s2v_config(&w2v, model, target, memory) * * eff_sentences, eff_words = populate_base_s2v_config( # <<<<<<<<<<<<<< * &w2v, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 261, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 312, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -2616,7 +2789,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 261, __pyx_L1_error) + __PYX_ERR(0, 312, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -2629,15 +2802,15 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_5); __Pyx_INCREF(__pyx_t_6); #else - __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 261, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 312, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); - __pyx_t_6 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 261, __pyx_L1_error) + __pyx_t_6 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 312, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 261, __pyx_L1_error) + __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 312, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_8 = Py_TYPE(__pyx_t_7)->tp_iternext; @@ -2645,7 +2818,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_5); index = 1; __pyx_t_6 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_6)) goto __pyx_L4_unpacking_failed; __Pyx_GOTREF(__pyx_t_6); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 261, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 312, __pyx_L1_error) __pyx_t_8 = NULL; __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; goto __pyx_L5_unpacking_done; @@ -2653,31 +2826,31 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; __pyx_t_8 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 261, __pyx_L1_error) + __PYX_ERR(0, 312, __pyx_L1_error) __pyx_L5_unpacking_done:; } - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 261, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 312, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 261, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 312, __pyx_L1_error) __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_v_eff_sentences = __pyx_t_2; __pyx_v_eff_words = __pyx_t_9; - /* "pooling_inner.pyx":267 + /* "pooling_inner.pyx":318 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: * compute_base_sentence_pooling( */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 267, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 318, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 267, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 318, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_3 = ((!__pyx_t_4) != 0); if (__pyx_t_3) { - /* "pooling_inner.pyx":268 + /* "pooling_inner.pyx":319 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -2692,7 +2865,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":269 + /* "pooling_inner.pyx":320 * if not model.hierarchical: * with nogil: * compute_base_sentence_pooling( # <<<<<<<<<<<<<< @@ -2702,7 +2875,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_base_sentence_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":268 + /* "pooling_inner.pyx":319 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -2721,7 +2894,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } } - /* "pooling_inner.pyx":267 + /* "pooling_inner.pyx":318 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< @@ -2731,7 +2904,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L6; } - /* "pooling_inner.pyx":274 + /* "pooling_inner.pyx":325 * ) * else: * with nogil: # <<<<<<<<<<<<<< @@ -2747,7 +2920,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":275 + /* "pooling_inner.pyx":326 * else: * with nogil: * compute_base_sentence_hier_pooling( # <<<<<<<<<<<<<< @@ -2757,7 +2930,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences, __pyx_v_window_size); } - /* "pooling_inner.pyx":274 + /* "pooling_inner.pyx":325 * ) * else: * with nogil: # <<<<<<<<<<<<<< @@ -2778,7 +2951,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } __pyx_L6:; - /* "pooling_inner.pyx":258 + /* "pooling_inner.pyx":309 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< @@ -2788,7 +2961,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L3; } - /* "pooling_inner.pyx":281 + /* "pooling_inner.pyx":332 * ) * else: * init_ft_s2v_config(&ft, model, target, memory) # <<<<<<<<<<<<<< @@ -2796,31 +2969,31 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec * eff_sentences, eff_words = populate_ft_s2v_config( */ /*else*/ { - __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 281, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 332, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":285 + /* "pooling_inner.pyx":336 * eff_sentences, eff_words = populate_ft_s2v_config( * &ft, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 285, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 336, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_6 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 285, __pyx_L1_error) + __pyx_t_6 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 336, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":283 + /* "pooling_inner.pyx":334 * init_ft_s2v_config(&ft, model, target, memory) * * eff_sentences, eff_words = populate_ft_s2v_config( # <<<<<<<<<<<<<< * &ft, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_6, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 283, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_6, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 334, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -2829,7 +3002,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 283, __pyx_L1_error) + __PYX_ERR(0, 334, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -2842,15 +3015,15 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_6); __Pyx_INCREF(__pyx_t_5); #else - __pyx_t_6 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 283, __pyx_L1_error) + __pyx_t_6 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 334, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); - __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 283, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 334, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 283, __pyx_L1_error) + __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 334, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_8 = Py_TYPE(__pyx_t_7)->tp_iternext; @@ -2858,7 +3031,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_6); index = 1; __pyx_t_5 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_5)) goto __pyx_L13_unpacking_failed; __Pyx_GOTREF(__pyx_t_5); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 283, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 334, __pyx_L1_error) __pyx_t_8 = NULL; __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; goto __pyx_L14_unpacking_done; @@ -2866,31 +3039,31 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; __pyx_t_8 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 283, __pyx_L1_error) + __PYX_ERR(0, 334, __pyx_L1_error) __pyx_L14_unpacking_done:; } - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 283, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 334, __pyx_L1_error) __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 283, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 334, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; __pyx_v_eff_sentences = __pyx_t_9; __pyx_v_eff_words = __pyx_t_2; - /* "pooling_inner.pyx":289 + /* "pooling_inner.pyx":340 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: * compute_ft_sentence_pooling(&ft, eff_sentences) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 289, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 340, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 289, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 340, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_4 = ((!__pyx_t_3) != 0); if (__pyx_t_4) { - /* "pooling_inner.pyx":290 + /* "pooling_inner.pyx":341 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -2905,7 +3078,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":291 + /* "pooling_inner.pyx":342 * if not model.hierarchical: * with nogil: * compute_ft_sentence_pooling(&ft, eff_sentences) # <<<<<<<<<<<<<< @@ -2915,7 +3088,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_ft_sentence_pooling((&__pyx_v_ft), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":290 + /* "pooling_inner.pyx":341 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -2934,7 +3107,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } } - /* "pooling_inner.pyx":289 + /* "pooling_inner.pyx":340 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< @@ -2945,7 +3118,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } __pyx_L3:; - /* "pooling_inner.pyx":293 + /* "pooling_inner.pyx":344 * compute_ft_sentence_pooling(&ft, eff_sentences) * * return eff_sentences, eff_words # <<<<<<<<<<<<<< @@ -2953,11 +3126,11 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec * def init(): */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 293, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 344, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 293, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 344, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); - __pyx_t_6 = PyTuple_New(2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 293, __pyx_L1_error) + __pyx_t_6 = PyTuple_New(2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 344, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_6, 0, __pyx_t_1); @@ -2969,8 +3142,8 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_6 = 0; goto __pyx_L0; - /* "pooling_inner.pyx":223 - * + /* "pooling_inner.pyx":274 + * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, @@ -2991,7 +3164,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec return __pyx_r; } -/* "pooling_inner.pyx":295 +/* "pooling_inner.pyx":346 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -3019,7 +3192,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("init", 0); - /* "pooling_inner.pyx":296 + /* "pooling_inner.pyx":347 * * def init(): * return 1 # <<<<<<<<<<<<<< @@ -3031,7 +3204,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __pyx_r = __pyx_int_1; goto __pyx_L0; - /* "pooling_inner.pyx":295 + /* "pooling_inner.pyx":346 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -5555,7 +5728,7 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = { {0, 0, 0, 0, 0, 0, 0} }; static CYTHON_SMALL_CODE int __Pyx_InitCachedBuiltins(void) { - __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 92, __pyx_L1_error) + __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 96, __pyx_L1_error) __pyx_builtin_ValueError = __Pyx_GetBuiltinName(__pyx_n_s_ValueError); if (!__pyx_builtin_ValueError) __PYX_ERR(1, 272, __pyx_L1_error) __pyx_builtin_RuntimeError = __Pyx_GetBuiltinName(__pyx_n_s_RuntimeError); if (!__pyx_builtin_RuntimeError) __PYX_ERR(1, 856, __pyx_L1_error) __pyx_builtin_ImportError = __Pyx_GetBuiltinName(__pyx_n_s_ImportError); if (!__pyx_builtin_ImportError) __PYX_ERR(1, 1038, __pyx_L1_error) @@ -5645,26 +5818,26 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { __Pyx_GOTREF(__pyx_tuple__7); __Pyx_GIVEREF(__pyx_tuple__7); - /* "pooling_inner.pyx":223 - * + /* "pooling_inner.pyx":274 + * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_tuple__9 = PyTuple_Pack(9, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_window_size, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 223, __pyx_L1_error) + __pyx_tuple__9 = PyTuple_Pack(9, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_window_size, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 274, __pyx_L1_error) __Pyx_GOTREF(__pyx_tuple__9); __Pyx_GIVEREF(__pyx_tuple__9); - __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 9, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 223, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 223, __pyx_L1_error) + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 9, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 274, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 274, __pyx_L1_error) - /* "pooling_inner.pyx":295 + /* "pooling_inner.pyx":346 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 295, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 295, __pyx_L1_error) + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 346, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 346, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -6009,76 +6182,76 @@ if (!__Pyx_RefNanny) { if (PyDict_SetItem(__pyx_d, __pyx_n_s_np, __pyx_t_1) < 0) __PYX_ERR(0, 15, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":21 - * from libc.string cimport memset + /* "pooling_inner.pyx":22 + * from libc.stdio cimport printf * * import scipy.linalg.blas as fblas # <<<<<<<<<<<<<< * * from average_inner cimport ( */ - __pyx_t_1 = PyList_New(1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 21, __pyx_L1_error) + __pyx_t_1 = PyList_New(1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 22, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_INCREF(__pyx_n_s__8); __Pyx_GIVEREF(__pyx_n_s__8); PyList_SET_ITEM(__pyx_t_1, 0, __pyx_n_s__8); - __pyx_t_2 = __Pyx_Import(__pyx_n_s_scipy_linalg_blas, __pyx_t_1, -1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 21, __pyx_L1_error) + __pyx_t_2 = __Pyx_Import(__pyx_n_s_scipy_linalg_blas, __pyx_t_1, -1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 22, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_2) < 0) __PYX_ERR(0, 21, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_2) < 0) __PYX_ERR(0, 22, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":223 - * + /* "pooling_inner.pyx":274 + * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 223, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 274, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 223, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 274, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":295 + /* "pooling_inner.pyx":346 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 295, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 346, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 295, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 346, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":298 + /* "pooling_inner.pyx":349 * return 1 * * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 298, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 349, __pyx_L1_error) - /* "pooling_inner.pyx":299 + /* "pooling_inner.pyx":350 * * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 299, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 350, __pyx_L1_error) - /* "pooling_inner.pyx":300 + /* "pooling_inner.pyx":351 * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() # <<<<<<<<<<<<<< */ - __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 300, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 351, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 300, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 351, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 300, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 351, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; /* "pooling_inner.pyx":1 diff --git a/fse/models/pooling_inner.pyx b/fse/models/pooling_inner.pyx index b04419a..ff6fd69 100644 --- a/fse/models/pooling_inner.pyx +++ b/fse/models/pooling_inner.pyx @@ -17,6 +17,7 @@ import numpy as np cimport numpy as np from libc.string cimport memset +from libc.stdio cimport printf import scipy.linalg.blas as fblas @@ -40,17 +41,20 @@ from average_inner cimport ( DEF MAX_WORDS = 10000 DEF MAX_NGRAMS = 40 -cdef void sl_max_pool( +cdef void swrmax_pool( const int *N, - float *X, - const float *Y, + const float *alpha, + const float *X, + float *Y, ) nogil: - """ Performs single left max pooling op + """ Performs single right weighted max pooling op Parameters ---------- N : int * Vector size. + alpha : float * + Weighting applied to X. X : float * Left vector. Y : float * @@ -59,8 +63,8 @@ cdef void sl_max_pool( """ cdef int i for i from 0 <= i < N[0] by 1: - if X[i] < Y[i]: - X[i] = Y[i] + if (alpha[0] * X[i]) > Y[i]: + Y[i] = alpha[0] * X[i] cdef void compute_base_sentence_pooling( BaseSentenceVecsConfig *c, @@ -98,11 +102,13 @@ cdef void compute_base_sentence_pooling( sent_len += ONEF sent_row = c.sent_adresses[i] * size word_row = c.word_indices[i] * size + word_idx = c.word_indices[i] - sl_max_pool( + swrmax_pool( &size, + &c.word_weights[word_idx], + &c.word_vectors[word_row], &c.sentence_vectors[sent_row], - &c.word_vectors[word_row], ) # There's nothing to do here for many-to-one mappings @@ -120,20 +126,23 @@ cdef void compute_base_sentence_hier_pooling( A pointer to a fully initialized and populated struct. num_sentences : uINT_t The number of sentences used to train the model. + window_size : uINT_t + The local window size. Notes ----- This routine does not provide oov support. """ + cdef: int size = c.size - uINT_t sent_idx, sent_start, sent_end, sent_row + uINT_t sent_idx, sent_start, sent_end, sent_row, window_end uINT_t i, j, word_idx, word_row - REAL_t sent_len, inv_count + REAL_t sent_len, win_len, inv_count for sent_idx in range(num_sentences): sent_start = c.sentence_boundary[sent_idx] @@ -142,9 +151,50 @@ cdef void compute_base_sentence_hier_pooling( for i in range(sent_start, sent_end): sent_len += ONEF - sent_row = c.sent_adresses[i] * size - word_row = c.word_indices[i] * size + sent_row = c.sent_adresses[i] * size + if i + window_size > sent_end: + window_end = sent_end + else: + window_end = i + window_size + + # Compute the locally averaged window + win_len = ZEROF + memset(c.mem, 0, size * cython.sizeof(REAL_t)) + memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + for j in range(i, window_end): + win_len += ONEF + + word_row = c.word_indices[j] * size + word_idx = c.word_indices[j] + + saxpy( + &size, + &c.word_weights[word_idx], + &c.word_vectors[word_row], + &ONE, + c.mem, + &ONE + ) + + # Rescale for dynamic window size + if win_len > ZEROF: + inv_count = ONEF / win_len + saxpy( + &size, + &inv_count, + c.mem, + &ONE, + c.mem2, + &ONE + ) + + swrmax_pool( + &size, + &ONEF, + c.mem2, + &c.sentence_vectors[sent_row], + ) cdef void compute_ft_sentence_pooling( FTSentenceVecsConfig *c, @@ -192,10 +242,11 @@ cdef void compute_ft_sentence_pooling( if ngrams == 0: word_row = c.word_indices[i] * size - sl_max_pool( + swrmax_pool( &size, + &c.word_weights[word_idx], + &c.word_vectors[word_row], &c.sentence_vectors[sent_row], - &c.word_vectors[word_row], ) else: @@ -212,10 +263,11 @@ cdef void compute_ft_sentence_pooling( &ONE ) - sl_max_pool( + swrmax_pool( &size, - &c.sentence_vectors[sent_row], + &oov_weight, c.mem, + &c.sentence_vectors[sent_row], ) # There's nothing to do here for many-to-one mappings @@ -274,7 +326,7 @@ def train_pooling_cy( compute_base_sentence_hier_pooling( &w2v, eff_sentences, - window_size + window_size, ) else: init_ft_s2v_config(&ft, model, target, memory) diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index 9cc35dc..74399e2 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -47,6 +47,7 @@ class TestPoolingFunctions(unittest.TestCase): + def setUp(self): self.sentences = [ ["They", "admit"], @@ -256,6 +257,16 @@ def test_pooling_train_np_w2v_non_negative(self): mpool.train(self.sentences) self.assertTrue((mpool.sv.vectors >= 0).all()) + def test_hpooling_train_np_w2v_non_negative(self): + mpool = MaxPooling(W2V_R, hierarchical=True) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) + + def test_pooling_train_np_ft_non_negative(self): + mpool = MaxPooling(FT_R) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) + def test_hier_pooling_train_np_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() @@ -271,15 +282,23 @@ def test_hier_pooling_train_np_w2v(self): self.assertTrue((183 == self.model.sv[0]).all()) self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) - def test_hpooling_train_np_w2v_non_negative(self): - mpool = MaxPooling(W2V_R, hierarchical=True) - mpool.train(self.sentences) - self.assertTrue((mpool.sv.vectors >= 0).all()) + def test_hier_pooling_train_cy_w2v(self): + self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + mem = self.model._get_thread_working_mem() + + self.model.hierarchical = True + + from fse.models.pooling_inner import train_pooling_cy + + output = train_pooling_cy( + self.model, self.sentences, self.model.sv.vectors, mem + ) + self.model.hierarchical = False + + self.assertEqual((5, 14), output) + self.assertTrue((183 == self.model.sv[0]).all()) + self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) - def test_pooling_train_np_ft_non_negative(self): - mpool = MaxPooling(FT_R) - mpool.train(self.sentences) - self.assertTrue((mpool.sv.vectors >= 0).all()) def test_hier_pooling_train_np_ft(self): m = MaxPooling(FT) From c5291e10f2cd50c549cebfd8830dc8fa04a1dc83 Mon Sep 17 00:00:00 2001 From: OliverB Date: Mon, 2 Mar 2020 18:18:07 +0100 Subject: [PATCH 31/63] Renaming i,j,k --- fse/models/average_inner.c | 892 +++++++++++++++++------------------ fse/models/average_inner.pyx | 53 ++- 2 files changed, 483 insertions(+), 462 deletions(-) diff --git a/fse/models/average_inner.c b/fse/models/average_inner.c index 79e6301..7f309c5 100644 --- a/fse/models/average_inner.c +++ b/fse/models/average_inner.c @@ -1959,7 +1959,7 @@ static PyObject *__pyx_codeobj__10; static PyObject *__pyx_codeobj__11; /* Late includes */ -/* "average_inner.pyx":37 +/* "average_inner.pyx":38 * DEF MAX_NGRAMS = 40 * * cdef init_base_s2v_config( # <<<<<<<<<<<<<< @@ -1975,101 +1975,101 @@ static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13a PyObject *__pyx_t_3 = NULL; __Pyx_RefNannySetupContext("init_base_s2v_config", 0); - /* "average_inner.pyx":58 + /* "average_inner.pyx":59 * * """ * c[0].workers = model.workers # <<<<<<<<<<<<<< * c[0].size = model.sv.vector_size * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_workers); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 58, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_workers); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 59, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 58, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 59, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; (__pyx_v_c[0]).workers = __pyx_t_2; - /* "average_inner.pyx":59 + /* "average_inner.pyx":60 * """ * c[0].workers = model.workers * c[0].size = model.sv.vector_size # <<<<<<<<<<<<<< * * c[0].mem = (np.PyArray_DATA(memory[0])) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_sv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 59, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_sv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 60, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vector_size); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 59, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vector_size); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 60, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 59, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 60, __pyx_L1_error) __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; (__pyx_v_c[0]).size = __pyx_t_2; - /* "average_inner.pyx":61 + /* "average_inner.pyx":62 * c[0].size = model.sv.vector_size * * c[0].mem = (np.PyArray_DATA(memory[0])) # <<<<<<<<<<<<<< * c[0].mem2 = (np.PyArray_DATA(memory[2])) * */ - __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_memory, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 61, __pyx_L1_error) + __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_memory, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 62, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 61, __pyx_L1_error) + if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 62, __pyx_L1_error) (__pyx_v_c[0]).mem = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_3))); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - /* "average_inner.pyx":62 + /* "average_inner.pyx":63 * * c[0].mem = (np.PyArray_DATA(memory[0])) * c[0].mem2 = (np.PyArray_DATA(memory[2])) # <<<<<<<<<<<<<< * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors)) */ - __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 62, __pyx_L1_error) + __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 63, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 62, __pyx_L1_error) + if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 63, __pyx_L1_error) (__pyx_v_c[0]).mem2 = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_3))); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - /* "average_inner.pyx":64 + /* "average_inner.pyx":65 * c[0].mem2 = (np.PyArray_DATA(memory[2])) * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors)) # <<<<<<<<<<<<<< * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) * */ - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 64, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 65, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_vectors); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 64, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_vectors); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 65, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 64, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 65, __pyx_L1_error) (__pyx_v_c[0]).word_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":65 + /* "average_inner.pyx":66 * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors)) * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) # <<<<<<<<<<<<<< * * c[0].sentence_vectors = (np.PyArray_DATA(target)) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 65, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 66, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 65, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 66, __pyx_L1_error) (__pyx_v_c[0]).word_weights = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":67 + /* "average_inner.pyx":68 * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) * * c[0].sentence_vectors = (np.PyArray_DATA(target)) # <<<<<<<<<<<<<< * * cdef init_ft_s2v_config( */ - if (!(likely(((__pyx_v_target) == Py_None) || likely(__Pyx_TypeTest(__pyx_v_target, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 67, __pyx_L1_error) + if (!(likely(((__pyx_v_target) == Py_None) || likely(__Pyx_TypeTest(__pyx_v_target, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 68, __pyx_L1_error) (__pyx_v_c[0]).sentence_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_v_target))); - /* "average_inner.pyx":37 + /* "average_inner.pyx":38 * DEF MAX_NGRAMS = 40 * * cdef init_base_s2v_config( # <<<<<<<<<<<<<< @@ -2091,7 +2091,7 @@ static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13a return __pyx_r; } -/* "average_inner.pyx":69 +/* "average_inner.pyx":70 * c[0].sentence_vectors = (np.PyArray_DATA(target)) * * cdef init_ft_s2v_config( # <<<<<<<<<<<<<< @@ -2110,96 +2110,96 @@ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13ave __pyx_t_13average_inner_REAL_t __pyx_t_6; __Pyx_RefNannySetupContext("init_ft_s2v_config", 0); - /* "average_inner.pyx":91 + /* "average_inner.pyx":92 * """ * * c[0].workers = model.workers # <<<<<<<<<<<<<< * c[0].size = model.sv.vector_size * c[0].min_n = model.wv.min_n */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_workers); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 91, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_workers); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 92, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 91, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 92, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; (__pyx_v_c[0]).workers = __pyx_t_2; - /* "average_inner.pyx":92 + /* "average_inner.pyx":93 * * c[0].workers = model.workers * c[0].size = model.sv.vector_size # <<<<<<<<<<<<<< * c[0].min_n = model.wv.min_n * c[0].max_n = model.wv.max_n */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_sv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 92, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_sv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 93, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vector_size); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 92, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vector_size); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 93, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 92, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 93, __pyx_L1_error) __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; (__pyx_v_c[0]).size = __pyx_t_2; - /* "average_inner.pyx":93 + /* "average_inner.pyx":94 * c[0].workers = model.workers * c[0].size = model.sv.vector_size * c[0].min_n = model.wv.min_n # <<<<<<<<<<<<<< * c[0].max_n = model.wv.max_n * c[0].bucket = model.wv.bucket */ - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 93, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 94, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_min_n); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 93, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_min_n); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 94, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 93, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 94, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; (__pyx_v_c[0]).min_n = __pyx_t_2; - /* "average_inner.pyx":94 + /* "average_inner.pyx":95 * c[0].size = model.sv.vector_size * c[0].min_n = model.wv.min_n * c[0].max_n = model.wv.max_n # <<<<<<<<<<<<<< * c[0].bucket = model.wv.bucket * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 94, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 95, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_max_n); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 94, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_max_n); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 95, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 94, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_3); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 95, __pyx_L1_error) __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; (__pyx_v_c[0]).max_n = __pyx_t_2; - /* "average_inner.pyx":95 + /* "average_inner.pyx":96 * c[0].min_n = model.wv.min_n * c[0].max_n = model.wv.max_n * c[0].bucket = model.wv.bucket # <<<<<<<<<<<<<< * * c[0].oov_weight = np.max(model.word_weights) */ - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 95, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 96, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_bucket); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 95, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_bucket); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 96, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 95, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_int(__pyx_t_1); if (unlikely((__pyx_t_2 == (int)-1) && PyErr_Occurred())) __PYX_ERR(0, 96, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; (__pyx_v_c[0]).bucket = __pyx_t_2; - /* "average_inner.pyx":97 + /* "average_inner.pyx":98 * c[0].bucket = model.wv.bucket * * c[0].oov_weight = np.max(model.word_weights) # <<<<<<<<<<<<<< * * c[0].mem = (np.PyArray_DATA(memory[0])) */ - __Pyx_GetModuleGlobalName(__pyx_t_3, __pyx_n_s_np); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 97, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_3, __pyx_n_s_np); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 98, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_max); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 97, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_max); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 98, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 97, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 98, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); __pyx_t_5 = NULL; if (CYTHON_UNPACK_METHODS && unlikely(PyMethod_Check(__pyx_t_4))) { @@ -2214,52 +2214,52 @@ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13ave __pyx_t_1 = (__pyx_t_5) ? __Pyx_PyObject_Call2Args(__pyx_t_4, __pyx_t_5, __pyx_t_3) : __Pyx_PyObject_CallOneArg(__pyx_t_4, __pyx_t_3); __Pyx_XDECREF(__pyx_t_5); __pyx_t_5 = 0; __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 97, __pyx_L1_error) + if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 98, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; - __pyx_t_6 = __pyx_PyFloat_AsFloat(__pyx_t_1); if (unlikely((__pyx_t_6 == ((npy_float32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 97, __pyx_L1_error) + __pyx_t_6 = __pyx_PyFloat_AsFloat(__pyx_t_1); if (unlikely((__pyx_t_6 == ((npy_float32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 98, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; (__pyx_v_c[0]).oov_weight = ((__pyx_t_13average_inner_REAL_t)__pyx_t_6); - /* "average_inner.pyx":99 + /* "average_inner.pyx":100 * c[0].oov_weight = np.max(model.word_weights) * * c[0].mem = (np.PyArray_DATA(memory[0])) # <<<<<<<<<<<<<< * c[0].mem2 = (np.PyArray_DATA(memory[2])) * */ - __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 99, __pyx_L1_error) + __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 100, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 99, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 100, __pyx_L1_error) (__pyx_v_c[0]).mem = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":100 + /* "average_inner.pyx":101 * * c[0].mem = (np.PyArray_DATA(memory[0])) * c[0].mem2 = (np.PyArray_DATA(memory[2])) # <<<<<<<<<<<<<< * * memory[1].fill(ZERO) # Reset the ngram storage before filling the struct */ - __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 100, __pyx_L1_error) + __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 101, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 100, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 101, __pyx_L1_error) (__pyx_v_c[0]).mem2 = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":102 + /* "average_inner.pyx":103 * c[0].mem2 = (np.PyArray_DATA(memory[2])) * * memory[1].fill(ZERO) # Reset the ngram storage before filling the struct # <<<<<<<<<<<<<< * c[0].subwords_idx = (np.PyArray_DATA(memory[1])) * */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 102, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 103, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_4, __pyx_n_s_fill); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 102, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_4, __pyx_n_s_fill); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 103, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; - __pyx_t_4 = __Pyx_PyInt_From_int(__pyx_v_13average_inner_ZERO); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 102, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyInt_From_int(__pyx_v_13average_inner_ZERO); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 103, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __pyx_t_5 = NULL; if (CYTHON_UNPACK_METHODS && likely(PyMethod_Check(__pyx_t_3))) { @@ -2274,80 +2274,80 @@ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13ave __pyx_t_1 = (__pyx_t_5) ? __Pyx_PyObject_Call2Args(__pyx_t_3, __pyx_t_5, __pyx_t_4) : __Pyx_PyObject_CallOneArg(__pyx_t_3, __pyx_t_4); __Pyx_XDECREF(__pyx_t_5); __pyx_t_5 = 0; __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; - if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 102, __pyx_L1_error) + if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 103, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":103 + /* "average_inner.pyx":104 * * memory[1].fill(ZERO) # Reset the ngram storage before filling the struct * c[0].subwords_idx = (np.PyArray_DATA(memory[1])) # <<<<<<<<<<<<<< * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) */ - __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 103, __pyx_L1_error) + __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 104, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 103, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 104, __pyx_L1_error) (__pyx_v_c[0]).subwords_idx = ((__pyx_t_13average_inner_uINT_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":105 + /* "average_inner.pyx":106 * c[0].subwords_idx = (np.PyArray_DATA(memory[1])) * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) # <<<<<<<<<<<<<< * c[0].ngram_vectors = (np.PyArray_DATA(model.wv.vectors_ngrams)) * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 105, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 106, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vectors_vocab); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 105, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vectors_vocab); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 106, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 105, __pyx_L1_error) + if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 106, __pyx_L1_error) (__pyx_v_c[0]).word_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_3))); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - /* "average_inner.pyx":106 + /* "average_inner.pyx":107 * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) * c[0].ngram_vectors = (np.PyArray_DATA(model.wv.vectors_ngrams)) # <<<<<<<<<<<<<< * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) * */ - __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 106, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 107, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_vectors_ngrams); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 106, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_3, __pyx_n_s_vectors_ngrams); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 107, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 106, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 107, __pyx_L1_error) (__pyx_v_c[0]).ngram_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":107 + /* "average_inner.pyx":108 * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) * c[0].ngram_vectors = (np.PyArray_DATA(model.wv.vectors_ngrams)) * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) # <<<<<<<<<<<<<< * * c[0].sentence_vectors = (np.PyArray_DATA(target)) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 107, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_word_weights); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 108, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 107, __pyx_L1_error) + if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 108, __pyx_L1_error) (__pyx_v_c[0]).word_weights = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":109 + /* "average_inner.pyx":110 * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) * * c[0].sentence_vectors = (np.PyArray_DATA(target)) # <<<<<<<<<<<<<< * * cdef object populate_base_s2v_config( */ - if (!(likely(((__pyx_v_target) == Py_None) || likely(__Pyx_TypeTest(__pyx_v_target, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 109, __pyx_L1_error) + if (!(likely(((__pyx_v_target) == Py_None) || likely(__Pyx_TypeTest(__pyx_v_target, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 110, __pyx_L1_error) (__pyx_v_c[0]).sentence_vectors = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_v_target))); - /* "average_inner.pyx":69 + /* "average_inner.pyx":70 * c[0].sentence_vectors = (np.PyArray_DATA(target)) * * cdef init_ft_s2v_config( # <<<<<<<<<<<<<< @@ -2371,7 +2371,7 @@ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13ave return __pyx_r; } -/* "average_inner.pyx":111 +/* "average_inner.pyx":112 * c[0].sentence_vectors = (np.PyArray_DATA(target)) * * cdef object populate_base_s2v_config( # <<<<<<<<<<<<<< @@ -2400,7 +2400,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_t_13average_inner_uINT_t __pyx_t_11; __Pyx_RefNannySetupContext("populate_base_s2v_config", 0); - /* "average_inner.pyx":139 + /* "average_inner.pyx":140 * """ * * cdef uINT_t eff_words = ZERO # Effective words encountered in a sentence # <<<<<<<<<<<<<< @@ -2409,7 +2409,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ __pyx_v_eff_words = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":140 + /* "average_inner.pyx":141 * * cdef uINT_t eff_words = ZERO # Effective words encountered in a sentence * cdef uINT_t eff_sents = ZERO # Effective sentences encountered # <<<<<<<<<<<<<< @@ -2418,7 +2418,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ __pyx_v_eff_sents = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":142 + /* "average_inner.pyx":143 * cdef uINT_t eff_sents = ZERO # Effective sentences encountered * * c.sentence_boundary[0] = ZERO # <<<<<<<<<<<<<< @@ -2427,7 +2427,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ (__pyx_v_c->sentence_boundary[0]) = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":144 + /* "average_inner.pyx":145 * c.sentence_boundary[0] = ZERO * * for obj in indexed_sentences: # <<<<<<<<<<<<<< @@ -2438,26 +2438,26 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_t_1 = __pyx_v_indexed_sentences; __Pyx_INCREF(__pyx_t_1); __pyx_t_2 = 0; __pyx_t_3 = NULL; } else { - __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 144, __pyx_L1_error) + __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 145, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext; if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 144, __pyx_L1_error) + __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext; if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 145, __pyx_L1_error) } for (;;) { if (likely(!__pyx_t_3)) { if (likely(PyList_CheckExact(__pyx_t_1))) { if (__pyx_t_2 >= PyList_GET_SIZE(__pyx_t_1)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 144, __pyx_L1_error) + __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 145, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 144, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 145, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } else { if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_1)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 144, __pyx_L1_error) + __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 145, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 144, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 145, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } @@ -2467,7 +2467,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t PyObject* exc_type = PyErr_Occurred(); if (exc_type) { if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); - else __PYX_ERR(0, 144, __pyx_L1_error) + else __PYX_ERR(0, 145, __pyx_L1_error) } break; } @@ -2476,21 +2476,21 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __Pyx_XDECREF_SET(__pyx_v_obj, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":145 + /* "average_inner.pyx":146 * * for obj in indexed_sentences: * if not obj[0]: # <<<<<<<<<<<<<< * continue * for token in obj[0]: */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 145, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 146, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 145, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 146, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; __pyx_t_6 = ((!__pyx_t_5) != 0); if (__pyx_t_6) { - /* "average_inner.pyx":146 + /* "average_inner.pyx":147 * for obj in indexed_sentences: * if not obj[0]: * continue # <<<<<<<<<<<<<< @@ -2499,7 +2499,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ goto __pyx_L3_continue; - /* "average_inner.pyx":145 + /* "average_inner.pyx":146 * * for obj in indexed_sentences: * if not obj[0]: # <<<<<<<<<<<<<< @@ -2508,22 +2508,22 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ } - /* "average_inner.pyx":147 + /* "average_inner.pyx":148 * if not obj[0]: * continue * for token in obj[0]: # <<<<<<<<<<<<<< * word = vocab[token] if token in vocab else None # Vocab obj * if word is None: */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 147, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 148, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); if (likely(PyList_CheckExact(__pyx_t_4)) || PyTuple_CheckExact(__pyx_t_4)) { __pyx_t_7 = __pyx_t_4; __Pyx_INCREF(__pyx_t_7); __pyx_t_8 = 0; __pyx_t_9 = NULL; } else { - __pyx_t_8 = -1; __pyx_t_7 = PyObject_GetIter(__pyx_t_4); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 147, __pyx_L1_error) + __pyx_t_8 = -1; __pyx_t_7 = PyObject_GetIter(__pyx_t_4); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 148, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); - __pyx_t_9 = Py_TYPE(__pyx_t_7)->tp_iternext; if (unlikely(!__pyx_t_9)) __PYX_ERR(0, 147, __pyx_L1_error) + __pyx_t_9 = Py_TYPE(__pyx_t_7)->tp_iternext; if (unlikely(!__pyx_t_9)) __PYX_ERR(0, 148, __pyx_L1_error) } __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; for (;;) { @@ -2531,17 +2531,17 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t if (likely(PyList_CheckExact(__pyx_t_7))) { if (__pyx_t_8 >= PyList_GET_SIZE(__pyx_t_7)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyList_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 147, __pyx_L1_error) + __pyx_t_4 = PyList_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 148, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 147, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 148, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } else { if (__pyx_t_8 >= PyTuple_GET_SIZE(__pyx_t_7)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 147, __pyx_L1_error) + __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 148, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 147, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 148, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } @@ -2551,7 +2551,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t PyObject* exc_type = PyErr_Occurred(); if (exc_type) { if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); - else __PYX_ERR(0, 147, __pyx_L1_error) + else __PYX_ERR(0, 148, __pyx_L1_error) } break; } @@ -2560,16 +2560,16 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __Pyx_XDECREF_SET(__pyx_v_token, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":148 + /* "average_inner.pyx":149 * continue * for token in obj[0]: * word = vocab[token] if token in vocab else None # Vocab obj # <<<<<<<<<<<<<< * if word is None: * continue */ - __pyx_t_6 = (__Pyx_PySequence_ContainsTF(__pyx_v_token, __pyx_v_vocab, Py_EQ)); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(0, 148, __pyx_L1_error) + __pyx_t_6 = (__Pyx_PySequence_ContainsTF(__pyx_v_token, __pyx_v_vocab, Py_EQ)); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(0, 149, __pyx_L1_error) if ((__pyx_t_6 != 0)) { - __pyx_t_10 = __Pyx_PyObject_GetItem(__pyx_v_vocab, __pyx_v_token); if (unlikely(!__pyx_t_10)) __PYX_ERR(0, 148, __pyx_L1_error) + __pyx_t_10 = __Pyx_PyObject_GetItem(__pyx_v_vocab, __pyx_v_token); if (unlikely(!__pyx_t_10)) __PYX_ERR(0, 149, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_10); __pyx_t_4 = __pyx_t_10; __pyx_t_10 = 0; @@ -2580,7 +2580,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __Pyx_XDECREF_SET(__pyx_v_word, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":149 + /* "average_inner.pyx":150 * for token in obj[0]: * word = vocab[token] if token in vocab else None # Vocab obj * if word is None: # <<<<<<<<<<<<<< @@ -2591,7 +2591,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_t_5 = (__pyx_t_6 != 0); if (__pyx_t_5) { - /* "average_inner.pyx":150 + /* "average_inner.pyx":151 * word = vocab[token] if token in vocab else None # Vocab obj * if word is None: * continue # <<<<<<<<<<<<<< @@ -2600,7 +2600,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ goto __pyx_L6_continue; - /* "average_inner.pyx":149 + /* "average_inner.pyx":150 * for token in obj[0]: * word = vocab[token] if token in vocab else None # Vocab obj * if word is None: # <<<<<<<<<<<<<< @@ -2609,33 +2609,33 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ } - /* "average_inner.pyx":151 + /* "average_inner.pyx":152 * if word is None: * continue * c.word_indices[eff_words] = word.index # <<<<<<<<<<<<<< * c.sent_adresses[eff_words] = obj[1] * */ - __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_v_word, __pyx_n_s_index); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 151, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_v_word, __pyx_n_s_index); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 152, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_11 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_11 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 151, __pyx_L1_error) + __pyx_t_11 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_11 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 152, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; (__pyx_v_c->word_indices[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_11); - /* "average_inner.pyx":152 + /* "average_inner.pyx":153 * continue * c.word_indices[eff_words] = word.index * c.sent_adresses[eff_words] = obj[1] # <<<<<<<<<<<<<< * * eff_words += ONE */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 152, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 153, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_11 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_11 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 152, __pyx_L1_error) + __pyx_t_11 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_11 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 153, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; (__pyx_v_c->sent_adresses[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_11); - /* "average_inner.pyx":154 + /* "average_inner.pyx":155 * c.sent_adresses[eff_words] = obj[1] * * eff_words += ONE # <<<<<<<<<<<<<< @@ -2644,7 +2644,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ __pyx_v_eff_words = (__pyx_v_eff_words + __pyx_v_13average_inner_ONE); - /* "average_inner.pyx":155 + /* "average_inner.pyx":156 * * eff_words += ONE * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -2654,7 +2654,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_t_5 = ((__pyx_v_eff_words == 0x2710) != 0); if (__pyx_t_5) { - /* "average_inner.pyx":156 + /* "average_inner.pyx":157 * eff_words += ONE * if eff_words == MAX_WORDS: * break # <<<<<<<<<<<<<< @@ -2663,7 +2663,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ goto __pyx_L7_break; - /* "average_inner.pyx":155 + /* "average_inner.pyx":156 * * eff_words += ONE * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -2672,7 +2672,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ } - /* "average_inner.pyx":147 + /* "average_inner.pyx":148 * if not obj[0]: * continue * for token in obj[0]: # <<<<<<<<<<<<<< @@ -2684,7 +2684,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_L7_break:; __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; - /* "average_inner.pyx":157 + /* "average_inner.pyx":158 * if eff_words == MAX_WORDS: * break * eff_sents += 1 # <<<<<<<<<<<<<< @@ -2693,7 +2693,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ __pyx_v_eff_sents = (__pyx_v_eff_sents + 1); - /* "average_inner.pyx":158 + /* "average_inner.pyx":159 * break * eff_sents += 1 * c.sentence_boundary[eff_sents] = eff_words # <<<<<<<<<<<<<< @@ -2702,7 +2702,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ (__pyx_v_c->sentence_boundary[__pyx_v_eff_sents]) = __pyx_v_eff_words; - /* "average_inner.pyx":160 + /* "average_inner.pyx":161 * c.sentence_boundary[eff_sents] = eff_words * * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -2712,7 +2712,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_t_5 = ((__pyx_v_eff_words == 0x2710) != 0); if (__pyx_t_5) { - /* "average_inner.pyx":161 + /* "average_inner.pyx":162 * * if eff_words == MAX_WORDS: * break # <<<<<<<<<<<<<< @@ -2721,7 +2721,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ goto __pyx_L4_break; - /* "average_inner.pyx":160 + /* "average_inner.pyx":161 * c.sentence_boundary[eff_sents] = eff_words * * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -2730,7 +2730,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t */ } - /* "average_inner.pyx":144 + /* "average_inner.pyx":145 * c.sentence_boundary[0] = ZERO * * for obj in indexed_sentences: # <<<<<<<<<<<<<< @@ -2742,7 +2742,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_L4_break:; __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":163 + /* "average_inner.pyx":164 * break * * return eff_sents, eff_words # <<<<<<<<<<<<<< @@ -2750,11 +2750,11 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t * cdef object populate_ft_s2v_config( */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sents); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 163, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sents); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 164, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_7 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 163, __pyx_L1_error) + __pyx_t_7 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 164, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); - __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 163, __pyx_L1_error) + __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 164, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_1); @@ -2766,7 +2766,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t __pyx_t_4 = 0; goto __pyx_L0; - /* "average_inner.pyx":111 + /* "average_inner.pyx":112 * c[0].sentence_vectors = (np.PyArray_DATA(target)) * * cdef object populate_base_s2v_config( # <<<<<<<<<<<<<< @@ -2791,7 +2791,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t return __pyx_r; } -/* "average_inner.pyx":165 +/* "average_inner.pyx":166 * return eff_sents, eff_words * * cdef object populate_ft_s2v_config( # <<<<<<<<<<<<<< @@ -2834,7 +2834,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 Py_ssize_t __pyx_t_20; __Pyx_RefNannySetupContext("populate_ft_s2v_config", 0); - /* "average_inner.pyx":193 + /* "average_inner.pyx":194 * """ * * cdef uINT_t eff_words = ZERO # Effective words encountered in a sentence # <<<<<<<<<<<<<< @@ -2843,7 +2843,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ __pyx_v_eff_words = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":194 + /* "average_inner.pyx":195 * * cdef uINT_t eff_words = ZERO # Effective words encountered in a sentence * cdef uINT_t eff_sents = ZERO # Effective sentences encountered # <<<<<<<<<<<<<< @@ -2852,7 +2852,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ __pyx_v_eff_sents = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":196 + /* "average_inner.pyx":197 * cdef uINT_t eff_sents = ZERO # Effective sentences encountered * * c.sentence_boundary[0] = ZERO # <<<<<<<<<<<<<< @@ -2861,7 +2861,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ (__pyx_v_c->sentence_boundary[0]) = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":198 + /* "average_inner.pyx":199 * c.sentence_boundary[0] = ZERO * * for obj in indexed_sentences: # <<<<<<<<<<<<<< @@ -2872,26 +2872,26 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __pyx_t_1 = __pyx_v_indexed_sentences; __Pyx_INCREF(__pyx_t_1); __pyx_t_2 = 0; __pyx_t_3 = NULL; } else { - __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 198, __pyx_L1_error) + __pyx_t_2 = -1; __pyx_t_1 = PyObject_GetIter(__pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 199, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext; if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 198, __pyx_L1_error) + __pyx_t_3 = Py_TYPE(__pyx_t_1)->tp_iternext; if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 199, __pyx_L1_error) } for (;;) { if (likely(!__pyx_t_3)) { if (likely(PyList_CheckExact(__pyx_t_1))) { if (__pyx_t_2 >= PyList_GET_SIZE(__pyx_t_1)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 198, __pyx_L1_error) + __pyx_t_4 = PyList_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 199, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 198, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 199, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } else { if (__pyx_t_2 >= PyTuple_GET_SIZE(__pyx_t_1)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 198, __pyx_L1_error) + __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_1, __pyx_t_2); __Pyx_INCREF(__pyx_t_4); __pyx_t_2++; if (unlikely(0 < 0)) __PYX_ERR(0, 199, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 198, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_1, __pyx_t_2); __pyx_t_2++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 199, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } @@ -2901,7 +2901,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 PyObject* exc_type = PyErr_Occurred(); if (exc_type) { if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); - else __PYX_ERR(0, 198, __pyx_L1_error) + else __PYX_ERR(0, 199, __pyx_L1_error) } break; } @@ -2910,21 +2910,21 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __Pyx_XDECREF_SET(__pyx_v_obj, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":199 + /* "average_inner.pyx":200 * * for obj in indexed_sentences: * if not obj[0]: # <<<<<<<<<<<<<< * continue * for token in obj[0]: */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 199, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 200, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 199, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 200, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; __pyx_t_6 = ((!__pyx_t_5) != 0); if (__pyx_t_6) { - /* "average_inner.pyx":200 + /* "average_inner.pyx":201 * for obj in indexed_sentences: * if not obj[0]: * continue # <<<<<<<<<<<<<< @@ -2933,7 +2933,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ goto __pyx_L3_continue; - /* "average_inner.pyx":199 + /* "average_inner.pyx":200 * * for obj in indexed_sentences: * if not obj[0]: # <<<<<<<<<<<<<< @@ -2942,22 +2942,22 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ } - /* "average_inner.pyx":201 + /* "average_inner.pyx":202 * if not obj[0]: * continue * for token in obj[0]: # <<<<<<<<<<<<<< * c.sent_adresses[eff_words] = obj[1] * if token in vocab: */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 201, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 202, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); if (likely(PyList_CheckExact(__pyx_t_4)) || PyTuple_CheckExact(__pyx_t_4)) { __pyx_t_7 = __pyx_t_4; __Pyx_INCREF(__pyx_t_7); __pyx_t_8 = 0; __pyx_t_9 = NULL; } else { - __pyx_t_8 = -1; __pyx_t_7 = PyObject_GetIter(__pyx_t_4); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 201, __pyx_L1_error) + __pyx_t_8 = -1; __pyx_t_7 = PyObject_GetIter(__pyx_t_4); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 202, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); - __pyx_t_9 = Py_TYPE(__pyx_t_7)->tp_iternext; if (unlikely(!__pyx_t_9)) __PYX_ERR(0, 201, __pyx_L1_error) + __pyx_t_9 = Py_TYPE(__pyx_t_7)->tp_iternext; if (unlikely(!__pyx_t_9)) __PYX_ERR(0, 202, __pyx_L1_error) } __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; for (;;) { @@ -2965,17 +2965,17 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 if (likely(PyList_CheckExact(__pyx_t_7))) { if (__pyx_t_8 >= PyList_GET_SIZE(__pyx_t_7)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyList_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 201, __pyx_L1_error) + __pyx_t_4 = PyList_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 202, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 201, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 202, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } else { if (__pyx_t_8 >= PyTuple_GET_SIZE(__pyx_t_7)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 201, __pyx_L1_error) + __pyx_t_4 = PyTuple_GET_ITEM(__pyx_t_7, __pyx_t_8); __Pyx_INCREF(__pyx_t_4); __pyx_t_8++; if (unlikely(0 < 0)) __PYX_ERR(0, 202, __pyx_L1_error) #else - __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 201, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(__pyx_t_7, __pyx_t_8); __pyx_t_8++; if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 202, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif } @@ -2985,7 +2985,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 PyObject* exc_type = PyErr_Occurred(); if (exc_type) { if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); - else __PYX_ERR(0, 201, __pyx_L1_error) + else __PYX_ERR(0, 202, __pyx_L1_error) } break; } @@ -2994,56 +2994,56 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __Pyx_XDECREF_SET(__pyx_v_token, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":202 + /* "average_inner.pyx":203 * continue * for token in obj[0]: * c.sent_adresses[eff_words] = obj[1] # <<<<<<<<<<<<<< * if token in vocab: * # In Vocabulary */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 202, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_obj, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 203, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 202, __pyx_L1_error) + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 203, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; (__pyx_v_c->sent_adresses[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_10); - /* "average_inner.pyx":203 + /* "average_inner.pyx":204 * for token in obj[0]: * c.sent_adresses[eff_words] = obj[1] * if token in vocab: # <<<<<<<<<<<<<< * # In Vocabulary * word = vocab[token] */ - __pyx_t_6 = (__Pyx_PySequence_ContainsTF(__pyx_v_token, __pyx_v_vocab, Py_EQ)); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(0, 203, __pyx_L1_error) + __pyx_t_6 = (__Pyx_PySequence_ContainsTF(__pyx_v_token, __pyx_v_vocab, Py_EQ)); if (unlikely(__pyx_t_6 < 0)) __PYX_ERR(0, 204, __pyx_L1_error) __pyx_t_5 = (__pyx_t_6 != 0); if (__pyx_t_5) { - /* "average_inner.pyx":205 + /* "average_inner.pyx":206 * if token in vocab: * # In Vocabulary * word = vocab[token] # <<<<<<<<<<<<<< * c.word_indices[eff_words] = word.index * c.subwords_idx_len[eff_words] = ZERO */ - __pyx_t_4 = __Pyx_PyObject_GetItem(__pyx_v_vocab, __pyx_v_token); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 205, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetItem(__pyx_v_vocab, __pyx_v_token); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 206, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_XDECREF_SET(__pyx_v_word, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":206 + /* "average_inner.pyx":207 * # In Vocabulary * word = vocab[token] * c.word_indices[eff_words] = word.index # <<<<<<<<<<<<<< * c.subwords_idx_len[eff_words] = ZERO * else: */ - __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_v_word, __pyx_n_s_index); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 206, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_v_word, __pyx_n_s_index); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 207, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 206, __pyx_L1_error) + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 207, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; (__pyx_v_c->word_indices[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_10); - /* "average_inner.pyx":207 + /* "average_inner.pyx":208 * word = vocab[token] * c.word_indices[eff_words] = word.index * c.subwords_idx_len[eff_words] = ZERO # <<<<<<<<<<<<<< @@ -3052,7 +3052,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ (__pyx_v_c->subwords_idx_len[__pyx_v_eff_words]) = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":203 + /* "average_inner.pyx":204 * for token in obj[0]: * c.sent_adresses[eff_words] = obj[1] * if token in vocab: # <<<<<<<<<<<<<< @@ -3062,7 +3062,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 goto __pyx_L8; } - /* "average_inner.pyx":210 + /* "average_inner.pyx":211 * else: * # OOV words --> write ngram indices to memory * c.word_indices[eff_words] = ZERO # <<<<<<<<<<<<<< @@ -3072,18 +3072,18 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 /*else*/ { (__pyx_v_c->word_indices[__pyx_v_eff_words]) = __pyx_v_13average_inner_ZERO; - /* "average_inner.pyx":212 + /* "average_inner.pyx":213 * c.word_indices[eff_words] = ZERO * * encoded_ngrams = compute_ngrams_bytes(token, c.min_n, c.max_n) # <<<<<<<<<<<<<< * hashes = [ft_hash_bytes(n) % c.bucket for n in encoded_ngrams] * */ - __Pyx_GetModuleGlobalName(__pyx_t_11, __pyx_n_s_compute_ngrams_bytes); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 212, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_11, __pyx_n_s_compute_ngrams_bytes); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_11); - __pyx_t_12 = __Pyx_PyInt_From_int(__pyx_v_c->min_n); if (unlikely(!__pyx_t_12)) __PYX_ERR(0, 212, __pyx_L1_error) + __pyx_t_12 = __Pyx_PyInt_From_int(__pyx_v_c->min_n); if (unlikely(!__pyx_t_12)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_12); - __pyx_t_13 = __Pyx_PyInt_From_int(__pyx_v_c->max_n); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 212, __pyx_L1_error) + __pyx_t_13 = __Pyx_PyInt_From_int(__pyx_v_c->max_n); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_13); __pyx_t_14 = NULL; __pyx_t_15 = 0; @@ -3100,7 +3100,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 #if CYTHON_FAST_PYCALL if (PyFunction_Check(__pyx_t_11)) { PyObject *__pyx_temp[4] = {__pyx_t_14, __pyx_v_token, __pyx_t_12, __pyx_t_13}; - __pyx_t_4 = __Pyx_PyFunction_FastCall(__pyx_t_11, __pyx_temp+1-__pyx_t_15, 3+__pyx_t_15); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 212, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyFunction_FastCall(__pyx_t_11, __pyx_temp+1-__pyx_t_15, 3+__pyx_t_15); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_XDECREF(__pyx_t_14); __pyx_t_14 = 0; __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0; @@ -3110,7 +3110,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 #if CYTHON_FAST_PYCCALL if (__Pyx_PyFastCFunction_Check(__pyx_t_11)) { PyObject *__pyx_temp[4] = {__pyx_t_14, __pyx_v_token, __pyx_t_12, __pyx_t_13}; - __pyx_t_4 = __Pyx_PyCFunction_FastCall(__pyx_t_11, __pyx_temp+1-__pyx_t_15, 3+__pyx_t_15); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 212, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyCFunction_FastCall(__pyx_t_11, __pyx_temp+1-__pyx_t_15, 3+__pyx_t_15); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_XDECREF(__pyx_t_14); __pyx_t_14 = 0; __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0; @@ -3118,7 +3118,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 } else #endif { - __pyx_t_16 = PyTuple_New(3+__pyx_t_15); if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 212, __pyx_L1_error) + __pyx_t_16 = PyTuple_New(3+__pyx_t_15); if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_16); if (__pyx_t_14) { __Pyx_GIVEREF(__pyx_t_14); PyTuple_SET_ITEM(__pyx_t_16, 0, __pyx_t_14); __pyx_t_14 = NULL; @@ -3132,7 +3132,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 PyTuple_SET_ITEM(__pyx_t_16, 2+__pyx_t_15, __pyx_t_13); __pyx_t_12 = 0; __pyx_t_13 = 0; - __pyx_t_4 = __Pyx_PyObject_Call(__pyx_t_11, __pyx_t_16, NULL); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 212, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_Call(__pyx_t_11, __pyx_t_16, NULL); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 213, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_16); __pyx_t_16 = 0; } @@ -3140,39 +3140,39 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __Pyx_XDECREF_SET(__pyx_v_encoded_ngrams, __pyx_t_4); __pyx_t_4 = 0; - /* "average_inner.pyx":213 + /* "average_inner.pyx":214 * * encoded_ngrams = compute_ngrams_bytes(token, c.min_n, c.max_n) * hashes = [ft_hash_bytes(n) % c.bucket for n in encoded_ngrams] # <<<<<<<<<<<<<< * * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) */ - __pyx_t_4 = PyList_New(0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 213, __pyx_L1_error) + __pyx_t_4 = PyList_New(0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 214, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); if (likely(PyList_CheckExact(__pyx_v_encoded_ngrams)) || PyTuple_CheckExact(__pyx_v_encoded_ngrams)) { __pyx_t_11 = __pyx_v_encoded_ngrams; __Pyx_INCREF(__pyx_t_11); __pyx_t_17 = 0; __pyx_t_18 = NULL; } else { - __pyx_t_17 = -1; __pyx_t_11 = PyObject_GetIter(__pyx_v_encoded_ngrams); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 213, __pyx_L1_error) + __pyx_t_17 = -1; __pyx_t_11 = PyObject_GetIter(__pyx_v_encoded_ngrams); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 214, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_11); - __pyx_t_18 = Py_TYPE(__pyx_t_11)->tp_iternext; if (unlikely(!__pyx_t_18)) __PYX_ERR(0, 213, __pyx_L1_error) + __pyx_t_18 = Py_TYPE(__pyx_t_11)->tp_iternext; if (unlikely(!__pyx_t_18)) __PYX_ERR(0, 214, __pyx_L1_error) } for (;;) { if (likely(!__pyx_t_18)) { if (likely(PyList_CheckExact(__pyx_t_11))) { if (__pyx_t_17 >= PyList_GET_SIZE(__pyx_t_11)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_16 = PyList_GET_ITEM(__pyx_t_11, __pyx_t_17); __Pyx_INCREF(__pyx_t_16); __pyx_t_17++; if (unlikely(0 < 0)) __PYX_ERR(0, 213, __pyx_L1_error) + __pyx_t_16 = PyList_GET_ITEM(__pyx_t_11, __pyx_t_17); __Pyx_INCREF(__pyx_t_16); __pyx_t_17++; if (unlikely(0 < 0)) __PYX_ERR(0, 214, __pyx_L1_error) #else - __pyx_t_16 = PySequence_ITEM(__pyx_t_11, __pyx_t_17); __pyx_t_17++; if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 213, __pyx_L1_error) + __pyx_t_16 = PySequence_ITEM(__pyx_t_11, __pyx_t_17); __pyx_t_17++; if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 214, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_16); #endif } else { if (__pyx_t_17 >= PyTuple_GET_SIZE(__pyx_t_11)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_16 = PyTuple_GET_ITEM(__pyx_t_11, __pyx_t_17); __Pyx_INCREF(__pyx_t_16); __pyx_t_17++; if (unlikely(0 < 0)) __PYX_ERR(0, 213, __pyx_L1_error) + __pyx_t_16 = PyTuple_GET_ITEM(__pyx_t_11, __pyx_t_17); __Pyx_INCREF(__pyx_t_16); __pyx_t_17++; if (unlikely(0 < 0)) __PYX_ERR(0, 214, __pyx_L1_error) #else - __pyx_t_16 = PySequence_ITEM(__pyx_t_11, __pyx_t_17); __pyx_t_17++; if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 213, __pyx_L1_error) + __pyx_t_16 = PySequence_ITEM(__pyx_t_11, __pyx_t_17); __pyx_t_17++; if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 214, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_16); #endif } @@ -3182,7 +3182,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 PyObject* exc_type = PyErr_Occurred(); if (exc_type) { if (likely(__Pyx_PyErr_GivenExceptionMatches(exc_type, PyExc_StopIteration))) PyErr_Clear(); - else __PYX_ERR(0, 213, __pyx_L1_error) + else __PYX_ERR(0, 214, __pyx_L1_error) } break; } @@ -3190,7 +3190,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 } __Pyx_XDECREF_SET(__pyx_v_n, __pyx_t_16); __pyx_t_16 = 0; - __Pyx_GetModuleGlobalName(__pyx_t_13, __pyx_n_s_ft_hash_bytes); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 213, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_13, __pyx_n_s_ft_hash_bytes); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 214, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_13); __pyx_t_12 = NULL; if (CYTHON_UNPACK_METHODS && unlikely(PyMethod_Check(__pyx_t_13))) { @@ -3204,23 +3204,23 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 } __pyx_t_16 = (__pyx_t_12) ? __Pyx_PyObject_Call2Args(__pyx_t_13, __pyx_t_12, __pyx_v_n) : __Pyx_PyObject_CallOneArg(__pyx_t_13, __pyx_v_n); __Pyx_XDECREF(__pyx_t_12); __pyx_t_12 = 0; - if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 213, __pyx_L1_error) + if (unlikely(!__pyx_t_16)) __PYX_ERR(0, 214, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_16); __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0; - __pyx_t_13 = __Pyx_PyInt_From_int(__pyx_v_c->bucket); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 213, __pyx_L1_error) + __pyx_t_13 = __Pyx_PyInt_From_int(__pyx_v_c->bucket); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 214, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_13); - __pyx_t_12 = PyNumber_Remainder(__pyx_t_16, __pyx_t_13); if (unlikely(!__pyx_t_12)) __PYX_ERR(0, 213, __pyx_L1_error) + __pyx_t_12 = PyNumber_Remainder(__pyx_t_16, __pyx_t_13); if (unlikely(!__pyx_t_12)) __PYX_ERR(0, 214, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_12); __Pyx_DECREF(__pyx_t_16); __pyx_t_16 = 0; __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0; - if (unlikely(__Pyx_ListComp_Append(__pyx_t_4, (PyObject*)__pyx_t_12))) __PYX_ERR(0, 213, __pyx_L1_error) + if (unlikely(__Pyx_ListComp_Append(__pyx_t_4, (PyObject*)__pyx_t_12))) __PYX_ERR(0, 214, __pyx_L1_error) __Pyx_DECREF(__pyx_t_12); __pyx_t_12 = 0; } __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 0; __Pyx_XDECREF_SET(__pyx_v_hashes, ((PyObject*)__pyx_t_4)); __pyx_t_4 = 0; - /* "average_inner.pyx":215 + /* "average_inner.pyx":216 * hashes = [ft_hash_bytes(n) % c.bucket for n in encoded_ngrams] * * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) # <<<<<<<<<<<<<< @@ -3228,7 +3228,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 * c.subwords_idx[(eff_words * MAX_NGRAMS) + i] = h */ __pyx_t_19 = 40; - __pyx_t_17 = PyObject_Length(__pyx_v_encoded_ngrams); if (unlikely(__pyx_t_17 == ((Py_ssize_t)-1))) __PYX_ERR(0, 215, __pyx_L1_error) + __pyx_t_17 = PyObject_Length(__pyx_v_encoded_ngrams); if (unlikely(__pyx_t_17 == ((Py_ssize_t)-1))) __PYX_ERR(0, 216, __pyx_L1_error) if (((__pyx_t_19 < __pyx_t_17) != 0)) { __pyx_t_20 = __pyx_t_19; } else { @@ -3236,7 +3236,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 } (__pyx_v_c->subwords_idx_len[__pyx_v_eff_words]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_20); - /* "average_inner.pyx":216 + /* "average_inner.pyx":217 * * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) * for i, h in enumerate(hashes[:MAX_NGRAMS]): # <<<<<<<<<<<<<< @@ -3245,46 +3245,46 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ __Pyx_INCREF(__pyx_int_0); __pyx_t_4 = __pyx_int_0; - __pyx_t_11 = __Pyx_PyList_GetSlice(__pyx_v_hashes, 0, 40); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 216, __pyx_L1_error) + __pyx_t_11 = __Pyx_PyList_GetSlice(__pyx_v_hashes, 0, 40); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 217, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_11); __pyx_t_12 = __pyx_t_11; __Pyx_INCREF(__pyx_t_12); __pyx_t_20 = 0; __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 0; for (;;) { if (__pyx_t_20 >= PyList_GET_SIZE(__pyx_t_12)) break; #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS - __pyx_t_11 = PyList_GET_ITEM(__pyx_t_12, __pyx_t_20); __Pyx_INCREF(__pyx_t_11); __pyx_t_20++; if (unlikely(0 < 0)) __PYX_ERR(0, 216, __pyx_L1_error) + __pyx_t_11 = PyList_GET_ITEM(__pyx_t_12, __pyx_t_20); __Pyx_INCREF(__pyx_t_11); __pyx_t_20++; if (unlikely(0 < 0)) __PYX_ERR(0, 217, __pyx_L1_error) #else - __pyx_t_11 = PySequence_ITEM(__pyx_t_12, __pyx_t_20); __pyx_t_20++; if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 216, __pyx_L1_error) + __pyx_t_11 = PySequence_ITEM(__pyx_t_12, __pyx_t_20); __pyx_t_20++; if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 217, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_11); #endif __Pyx_XDECREF_SET(__pyx_v_h, __pyx_t_11); __pyx_t_11 = 0; __Pyx_INCREF(__pyx_t_4); __Pyx_XDECREF_SET(__pyx_v_i, __pyx_t_4); - __pyx_t_11 = __Pyx_PyInt_AddObjC(__pyx_t_4, __pyx_int_1, 1, 0, 0); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 216, __pyx_L1_error) + __pyx_t_11 = __Pyx_PyInt_AddObjC(__pyx_t_4, __pyx_int_1, 1, 0, 0); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 217, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_11); __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = __pyx_t_11; __pyx_t_11 = 0; - /* "average_inner.pyx":217 + /* "average_inner.pyx":218 * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) * for i, h in enumerate(hashes[:MAX_NGRAMS]): * c.subwords_idx[(eff_words * MAX_NGRAMS) + i] = h # <<<<<<<<<<<<<< * * eff_words += ONE */ - __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_v_h); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 217, __pyx_L1_error) - __pyx_t_11 = __Pyx_PyInt_From_long((__pyx_v_eff_words * 40)); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 217, __pyx_L1_error) + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_v_h); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 218, __pyx_L1_error) + __pyx_t_11 = __Pyx_PyInt_From_long((__pyx_v_eff_words * 40)); if (unlikely(!__pyx_t_11)) __PYX_ERR(0, 218, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_11); - __pyx_t_13 = PyNumber_Add(__pyx_t_11, __pyx_v_i); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 217, __pyx_L1_error) + __pyx_t_13 = PyNumber_Add(__pyx_t_11, __pyx_v_i); if (unlikely(!__pyx_t_13)) __PYX_ERR(0, 218, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_13); __Pyx_DECREF(__pyx_t_11); __pyx_t_11 = 0; - __pyx_t_17 = __Pyx_PyIndex_AsSsize_t(__pyx_t_13); if (unlikely((__pyx_t_17 == (Py_ssize_t)-1) && PyErr_Occurred())) __PYX_ERR(0, 217, __pyx_L1_error) + __pyx_t_17 = __Pyx_PyIndex_AsSsize_t(__pyx_t_13); if (unlikely((__pyx_t_17 == (Py_ssize_t)-1) && PyErr_Occurred())) __PYX_ERR(0, 218, __pyx_L1_error) __Pyx_DECREF(__pyx_t_13); __pyx_t_13 = 0; (__pyx_v_c->subwords_idx[__pyx_t_17]) = ((__pyx_t_13average_inner_uINT_t)__pyx_t_10); - /* "average_inner.pyx":216 + /* "average_inner.pyx":217 * * c.subwords_idx_len[eff_words] = min(len(encoded_ngrams), MAX_NGRAMS) * for i, h in enumerate(hashes[:MAX_NGRAMS]): # <<<<<<<<<<<<<< @@ -3297,7 +3297,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 } __pyx_L8:; - /* "average_inner.pyx":219 + /* "average_inner.pyx":220 * c.subwords_idx[(eff_words * MAX_NGRAMS) + i] = h * * eff_words += ONE # <<<<<<<<<<<<<< @@ -3306,7 +3306,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ __pyx_v_eff_words = (__pyx_v_eff_words + __pyx_v_13average_inner_ONE); - /* "average_inner.pyx":221 + /* "average_inner.pyx":222 * eff_words += ONE * * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -3316,7 +3316,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __pyx_t_5 = ((__pyx_v_eff_words == 0x2710) != 0); if (__pyx_t_5) { - /* "average_inner.pyx":222 + /* "average_inner.pyx":223 * * if eff_words == MAX_WORDS: * break # <<<<<<<<<<<<<< @@ -3325,7 +3325,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ goto __pyx_L7_break; - /* "average_inner.pyx":221 + /* "average_inner.pyx":222 * eff_words += ONE * * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -3334,7 +3334,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ } - /* "average_inner.pyx":201 + /* "average_inner.pyx":202 * if not obj[0]: * continue * for token in obj[0]: # <<<<<<<<<<<<<< @@ -3345,7 +3345,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __pyx_L7_break:; __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; - /* "average_inner.pyx":224 + /* "average_inner.pyx":225 * break * * eff_sents += 1 # <<<<<<<<<<<<<< @@ -3354,7 +3354,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ __pyx_v_eff_sents = (__pyx_v_eff_sents + 1); - /* "average_inner.pyx":225 + /* "average_inner.pyx":226 * * eff_sents += 1 * c.sentence_boundary[eff_sents] = eff_words # <<<<<<<<<<<<<< @@ -3363,7 +3363,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ (__pyx_v_c->sentence_boundary[__pyx_v_eff_sents]) = __pyx_v_eff_words; - /* "average_inner.pyx":227 + /* "average_inner.pyx":228 * c.sentence_boundary[eff_sents] = eff_words * * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -3373,7 +3373,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __pyx_t_5 = ((__pyx_v_eff_words == 0x2710) != 0); if (__pyx_t_5) { - /* "average_inner.pyx":228 + /* "average_inner.pyx":229 * * if eff_words == MAX_WORDS: * break # <<<<<<<<<<<<<< @@ -3382,7 +3382,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ goto __pyx_L4_break; - /* "average_inner.pyx":227 + /* "average_inner.pyx":228 * c.sentence_boundary[eff_sents] = eff_words * * if eff_words == MAX_WORDS: # <<<<<<<<<<<<<< @@ -3391,7 +3391,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 */ } - /* "average_inner.pyx":198 + /* "average_inner.pyx":199 * c.sentence_boundary[0] = ZERO * * for obj in indexed_sentences: # <<<<<<<<<<<<<< @@ -3403,7 +3403,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __pyx_L4_break:; __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":230 + /* "average_inner.pyx":231 * break * * return eff_sents, eff_words # <<<<<<<<<<<<<< @@ -3411,11 +3411,11 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 * cdef void compute_base_sentence_averages( */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sents); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 230, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sents); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 231, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_7 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 230, __pyx_L1_error) + __pyx_t_7 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 231, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); - __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 230, __pyx_L1_error) + __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 231, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_1); @@ -3427,7 +3427,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 __pyx_t_4 = 0; goto __pyx_L0; - /* "average_inner.pyx":165 + /* "average_inner.pyx":166 * return eff_sents, eff_words * * cdef object populate_ft_s2v_config( # <<<<<<<<<<<<<< @@ -3461,7 +3461,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 return __pyx_r; } -/* "average_inner.pyx":232 +/* "average_inner.pyx":233 * return eff_sents, eff_words * * cdef void compute_base_sentence_averages( # <<<<<<<<<<<<<< @@ -3475,7 +3475,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; __pyx_t_13average_inner_uINT_t __pyx_v_sent_end; __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; - __pyx_t_13average_inner_uINT_t __pyx_v_i; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_pos; __pyx_t_13average_inner_uINT_t __pyx_v_word_idx; __pyx_t_13average_inner_uINT_t __pyx_v_word_row; __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; @@ -3489,7 +3489,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_t_13average_inner_uINT_t __pyx_t_7; int __pyx_t_8; - /* "average_inner.pyx":251 + /* "average_inner.pyx":252 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -3499,7 +3499,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "average_inner.pyx":259 + /* "average_inner.pyx":260 * REAL_t sent_len, inv_count * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -3511,7 +3511,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) { __pyx_v_sent_idx = __pyx_t_4; - /* "average_inner.pyx":260 + /* "average_inner.pyx":261 * * for sent_idx in range(num_sentences): * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -3520,7 +3520,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "average_inner.pyx":262 + /* "average_inner.pyx":263 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -3529,7 +3529,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "average_inner.pyx":263 + /* "average_inner.pyx":264 * * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -3538,65 +3538,65 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "average_inner.pyx":264 + /* "average_inner.pyx":265 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< * - * for i in range(sent_start, sent_end): + * for sent_pos in range(sent_start, sent_end): */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "average_inner.pyx":266 + /* "average_inner.pyx":267 * sent_len = ZEROF * - * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< + * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< * sent_len += ONEF - * sent_row = c.sent_adresses[i] * size + * sent_row = c.sent_adresses[sent_pos] * size */ __pyx_t_5 = __pyx_v_sent_end; __pyx_t_6 = __pyx_t_5; for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { - __pyx_v_i = __pyx_t_7; + __pyx_v_sent_pos = __pyx_t_7; - /* "average_inner.pyx":267 + /* "average_inner.pyx":268 * - * for i in range(sent_start, sent_end): + * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< - * sent_row = c.sent_adresses[i] * size - * word_row = c.word_indices[i] * size + * sent_row = c.sent_adresses[sent_pos] * size + * word_row = c.word_indices[sent_pos] * size */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "average_inner.pyx":268 - * for i in range(sent_start, sent_end): + /* "average_inner.pyx":269 + * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF - * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< - * word_row = c.word_indices[i] * size - * word_idx = c.word_indices[i] + * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< + * word_row = c.word_indices[sent_pos] * size + * word_idx = c.word_indices[sent_pos] */ - __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); + __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); - /* "average_inner.pyx":269 + /* "average_inner.pyx":270 * sent_len += ONEF - * sent_row = c.sent_adresses[i] * size - * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< - * word_idx = c.word_indices[i] + * sent_row = c.sent_adresses[sent_pos] * size + * word_row = c.word_indices[sent_pos] * size # <<<<<<<<<<<<<< + * word_idx = c.word_indices[sent_pos] * */ - __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); + __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_sent_pos]) * __pyx_v_size); - /* "average_inner.pyx":270 - * sent_row = c.sent_adresses[i] * size - * word_row = c.word_indices[i] * size - * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< + /* "average_inner.pyx":271 + * sent_row = c.sent_adresses[sent_pos] * size + * word_row = c.word_indices[sent_pos] * size + * word_idx = c.word_indices[sent_pos] # <<<<<<<<<<<<<< * * saxpy( */ - __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); + __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_sent_pos]); - /* "average_inner.pyx":272 - * word_idx = c.word_indices[i] + /* "average_inner.pyx":273 + * word_idx = c.word_indices[sent_pos] * * saxpy( # <<<<<<<<<<<<<< * &size, @@ -3605,7 +3605,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); } - /* "average_inner.pyx":281 + /* "average_inner.pyx":282 * ) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< @@ -3615,7 +3615,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_t_8 = ((__pyx_v_sent_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_8) { - /* "average_inner.pyx":282 + /* "average_inner.pyx":283 * * if sent_len > ZEROF: * inv_count = ONEF / sent_len # <<<<<<<<<<<<<< @@ -3624,7 +3624,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_sent_len); - /* "average_inner.pyx":285 + /* "average_inner.pyx":286 * # If we perform the a*x on memory, the computation is compatible with many-to-one mappings * # because it doesn't rescale the overall result * saxpy( # <<<<<<<<<<<<<< @@ -3633,7 +3633,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&__pyx_v_13average_inner_ONE)); - /* "average_inner.pyx":281 + /* "average_inner.pyx":282 * ) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< @@ -3643,7 +3643,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ } } - /* "average_inner.pyx":232 + /* "average_inner.pyx":233 * return eff_sents, eff_words * * cdef void compute_base_sentence_averages( # <<<<<<<<<<<<<< @@ -3654,7 +3654,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ /* function exit code */ } -/* "average_inner.pyx":294 +/* "average_inner.pyx":295 * ) * * cdef void compute_ft_sentence_averages( # <<<<<<<<<<<<<< @@ -3670,8 +3670,8 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; __pyx_t_13average_inner_uINT_t __pyx_v_ngram_row; __pyx_t_13average_inner_uINT_t __pyx_v_ngrams; - __pyx_t_13average_inner_uINT_t __pyx_v_i; - __pyx_t_13average_inner_uINT_t __pyx_v_j; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_pos; + __pyx_t_13average_inner_uINT_t __pyx_v_ngram_pos; __pyx_t_13average_inner_uINT_t __pyx_v_word_idx; __pyx_t_13average_inner_uINT_t __pyx_v_word_row; __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; @@ -3691,7 +3691,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_13average_inner_uINT_t __pyx_t_11; __pyx_t_13average_inner_uINT_t __pyx_t_12; - /* "average_inner.pyx":313 + /* "average_inner.pyx":314 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -3701,7 +3701,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "average_inner.pyx":323 + /* "average_inner.pyx":324 * REAL_t sent_len * REAL_t inv_count, inv_ngram * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< @@ -3711,7 +3711,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_2 = __pyx_v_c->oov_weight; __pyx_v_oov_weight = __pyx_t_2; - /* "average_inner.pyx":326 + /* "average_inner.pyx":327 * * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -3723,7 +3723,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { __pyx_v_sent_idx = __pyx_t_5; - /* "average_inner.pyx":327 + /* "average_inner.pyx":328 * * for sent_idx in range(num_sentences): * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -3732,7 +3732,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "average_inner.pyx":328 + /* "average_inner.pyx":329 * for sent_idx in range(num_sentences): * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -3741,7 +3741,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "average_inner.pyx":329 + /* "average_inner.pyx":330 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -3750,138 +3750,138 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "average_inner.pyx":330 + /* "average_inner.pyx":331 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< * - * for i in range(sent_start, sent_end): + * for sent_pos in range(sent_start, sent_end): */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "average_inner.pyx":332 + /* "average_inner.pyx":333 * sent_len = ZEROF * - * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< + * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< * sent_len += ONEF - * sent_row = c.sent_adresses[i] * size + * sent_row = c.sent_adresses[sent_pos] * size */ __pyx_t_6 = __pyx_v_sent_end; __pyx_t_7 = __pyx_t_6; for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { - __pyx_v_i = __pyx_t_8; + __pyx_v_sent_pos = __pyx_t_8; - /* "average_inner.pyx":333 + /* "average_inner.pyx":334 * - * for i in range(sent_start, sent_end): + * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< - * sent_row = c.sent_adresses[i] * size + * sent_row = c.sent_adresses[sent_pos] * size * */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "average_inner.pyx":334 - * for i in range(sent_start, sent_end): + /* "average_inner.pyx":335 + * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF - * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< + * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< * - * word_idx = c.word_indices[i] + * word_idx = c.word_indices[sent_pos] */ - __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); + __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); - /* "average_inner.pyx":336 - * sent_row = c.sent_adresses[i] * size + /* "average_inner.pyx":337 + * sent_row = c.sent_adresses[sent_pos] * size * - * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< - * ngrams = c.subwords_idx_len[i] + * word_idx = c.word_indices[sent_pos] # <<<<<<<<<<<<<< + * ngrams = c.subwords_idx_len[sent_pos] * */ - __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); + __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_sent_pos]); - /* "average_inner.pyx":337 + /* "average_inner.pyx":338 * - * word_idx = c.word_indices[i] - * ngrams = c.subwords_idx_len[i] # <<<<<<<<<<<<<< + * word_idx = c.word_indices[sent_pos] + * ngrams = c.subwords_idx_len[sent_pos] # <<<<<<<<<<<<<< * * if ngrams == 0: */ - __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_i]); + __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_sent_pos]); - /* "average_inner.pyx":339 - * ngrams = c.subwords_idx_len[i] + /* "average_inner.pyx":340 + * ngrams = c.subwords_idx_len[sent_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< - * word_row = c.word_indices[i] * size - * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + * word_row = c.word_indices[sent_pos] * size + * saxpy( */ __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); if (__pyx_t_9) { - /* "average_inner.pyx":340 + /* "average_inner.pyx":341 * * if ngrams == 0: - * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< - * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) - * else: + * word_row = c.word_indices[sent_pos] * size # <<<<<<<<<<<<<< + * saxpy( + * &size, */ - __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); + __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_sent_pos]) * __pyx_v_size); - /* "average_inner.pyx":341 + /* "average_inner.pyx":342 * if ngrams == 0: - * word_row = c.word_indices[i] * size - * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) # <<<<<<<<<<<<<< - * else: - * inv_ngram = (ONEF / ngrams) * c.oov_weight + * word_row = c.word_indices[sent_pos] * size + * saxpy( # <<<<<<<<<<<<<< + * &size, + * &c.word_weights[word_idx], */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); - /* "average_inner.pyx":339 - * ngrams = c.subwords_idx_len[i] + /* "average_inner.pyx":340 + * ngrams = c.subwords_idx_len[sent_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< - * word_row = c.word_indices[i] * size - * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + * word_row = c.word_indices[sent_pos] * size + * saxpy( */ goto __pyx_L7; } - /* "average_inner.pyx":343 - * saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + /* "average_inner.pyx":351 + * ) * else: * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< - * for j in range(ngrams): - * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size + * for ngram_pos in range(ngrams): + * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size */ /*else*/ { __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); - /* "average_inner.pyx":344 + /* "average_inner.pyx":352 * else: * inv_ngram = (ONEF / ngrams) * c.oov_weight - * for j in range(ngrams): # <<<<<<<<<<<<<< - * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size - * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) + * for ngram_pos in range(ngrams): # <<<<<<<<<<<<<< + * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size + * saxpy( */ __pyx_t_10 = __pyx_v_ngrams; __pyx_t_11 = __pyx_t_10; for (__pyx_t_12 = 0; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { - __pyx_v_j = __pyx_t_12; + __pyx_v_ngram_pos = __pyx_t_12; - /* "average_inner.pyx":345 + /* "average_inner.pyx":353 * inv_ngram = (ONEF / ngrams) * c.oov_weight - * for j in range(ngrams): - * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size # <<<<<<<<<<<<<< - * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) - * + * for ngram_pos in range(ngrams): + * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size # <<<<<<<<<<<<<< + * saxpy( + * &size, */ - __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_i * 40) + __pyx_v_j)]) * __pyx_v_size); + __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_sent_pos * 40) + __pyx_v_ngram_pos)]) * __pyx_v_size); - /* "average_inner.pyx":346 - * for j in range(ngrams): - * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size - * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) # <<<<<<<<<<<<<< - * - * if sent_len > ZEROF: + /* "average_inner.pyx":354 + * for ngram_pos in range(ngrams): + * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size + * saxpy( # <<<<<<<<<<<<<< + * &size, + * &inv_ngram, */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_ngram), (&(__pyx_v_c->ngram_vectors[__pyx_v_ngram_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); } @@ -3889,45 +3889,45 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_L7:; } - /* "average_inner.pyx":348 - * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) + /* "average_inner.pyx":363 + * ) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< * inv_count = ONEF / sent_len - * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + * saxpy( */ __pyx_t_9 = ((__pyx_v_sent_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_9) { - /* "average_inner.pyx":349 + /* "average_inner.pyx":364 * * if sent_len > ZEROF: * inv_count = ONEF / sent_len # <<<<<<<<<<<<<< - * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) - * + * saxpy( + * &size, */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_sent_len); - /* "average_inner.pyx":350 + /* "average_inner.pyx":365 * if sent_len > ZEROF: * inv_count = ONEF / sent_len - * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) # <<<<<<<<<<<<<< - * - * def train_average_cy( + * saxpy( # <<<<<<<<<<<<<< + * &size, + * &inv_count, */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&__pyx_v_13average_inner_ONE)); - /* "average_inner.pyx":348 - * saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) + /* "average_inner.pyx":363 + * ) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< * inv_count = ONEF / sent_len - * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + * saxpy( */ } } - /* "average_inner.pyx":294 + /* "average_inner.pyx":295 * ) * * cdef void compute_ft_sentence_averages( # <<<<<<<<<<<<<< @@ -3938,8 +3938,8 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ /* function exit code */ } -/* "average_inner.pyx":352 - * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) +/* "average_inner.pyx":374 + * ) * * def train_average_cy( # <<<<<<<<<<<<<< * model, @@ -3985,23 +3985,23 @@ static PyObject *__pyx_pw_13average_inner_1train_average_cy(PyObject *__pyx_self case 1: if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 1); __PYX_ERR(0, 352, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 1); __PYX_ERR(0, 374, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 2: if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 2); __PYX_ERR(0, 352, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 2); __PYX_ERR(0, 374, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 3: if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 3); __PYX_ERR(0, 352, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 3); __PYX_ERR(0, 374, __pyx_L3_error) } } if (unlikely(kw_args > 0)) { - if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_average_cy") < 0)) __PYX_ERR(0, 352, __pyx_L3_error) + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_average_cy") < 0)) __PYX_ERR(0, 374, __pyx_L3_error) } } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { goto __pyx_L5_argtuple_error; @@ -4018,7 +4018,7 @@ static PyObject *__pyx_pw_13average_inner_1train_average_cy(PyObject *__pyx_self } goto __pyx_L4_argument_unpacking_done; __pyx_L5_argtuple_error:; - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 352, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 374, __pyx_L3_error) __pyx_L3_error:; __Pyx_AddTraceback("average_inner.train_average_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __Pyx_RefNannyFinishContext(); @@ -4049,7 +4049,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_t_13average_inner_uINT_t __pyx_t_9; __Pyx_RefNannySetupContext("train_average_cy", 0); - /* "average_inner.pyx":381 + /* "average_inner.pyx":403 * """ * * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< @@ -4058,7 +4058,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_sentences = 0; - /* "average_inner.pyx":382 + /* "average_inner.pyx":404 * * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< @@ -4067,52 +4067,52 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_words = 0; - /* "average_inner.pyx":386 + /* "average_inner.pyx":408 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< * init_base_s2v_config(&w2v, model, target, memory) * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 386, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 408, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 386, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 408, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_3 = ((!__pyx_t_2) != 0); if (__pyx_t_3) { - /* "average_inner.pyx":387 + /* "average_inner.pyx":409 * * if not model.is_ft: * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< * * eff_sentences, eff_words = populate_base_s2v_config( */ - __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 387, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 409, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":391 + /* "average_inner.pyx":413 * eff_sentences, eff_words = populate_base_s2v_config( * &w2v, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences, * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 391, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 413, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 391, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 413, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":389 + /* "average_inner.pyx":411 * init_base_s2v_config(&w2v, model, target, memory) * * eff_sentences, eff_words = populate_base_s2v_config( # <<<<<<<<<<<<<< * &w2v, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 389, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 411, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -4121,7 +4121,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 389, __pyx_L1_error) + __PYX_ERR(0, 411, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -4134,15 +4134,15 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_4); __Pyx_INCREF(__pyx_t_5); #else - __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 389, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 411, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 389, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 411, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 389, __pyx_L1_error) + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 411, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; @@ -4150,7 +4150,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_4); index = 1; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L4_unpacking_failed; __Pyx_GOTREF(__pyx_t_5); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 389, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 411, __pyx_L1_error) __pyx_t_7 = NULL; __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; goto __pyx_L5_unpacking_done; @@ -4158,17 +4158,17 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_t_7 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 389, __pyx_L1_error) + __PYX_ERR(0, 411, __pyx_L1_error) __pyx_L5_unpacking_done:; } - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 389, __pyx_L1_error) + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 411, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 389, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 411, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; __pyx_v_eff_sentences = __pyx_t_8; __pyx_v_eff_words = __pyx_t_9; - /* "average_inner.pyx":395 + /* "average_inner.pyx":417 * ) * * with nogil: # <<<<<<<<<<<<<< @@ -4183,7 +4183,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "average_inner.pyx":396 + /* "average_inner.pyx":418 * * with nogil: * compute_base_sentence_averages( # <<<<<<<<<<<<<< @@ -4193,7 +4193,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_f_13average_inner_compute_base_sentence_averages((&__pyx_v_w2v), __pyx_v_eff_sentences); } - /* "average_inner.pyx":395 + /* "average_inner.pyx":417 * ) * * with nogil: # <<<<<<<<<<<<<< @@ -4212,7 +4212,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec } } - /* "average_inner.pyx":386 + /* "average_inner.pyx":408 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< @@ -4222,7 +4222,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec goto __pyx_L3; } - /* "average_inner.pyx":401 + /* "average_inner.pyx":423 * ) * else: * init_ft_s2v_config( # <<<<<<<<<<<<<< @@ -4231,38 +4231,38 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec */ /*else*/ { - /* "average_inner.pyx":405 + /* "average_inner.pyx":427 * model, * target, * memory, # <<<<<<<<<<<<<< * ) * */ - __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 401, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 423, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":410 + /* "average_inner.pyx":432 * eff_sentences, eff_words = populate_ft_s2v_config( * &ft, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences, * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 410, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 432, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 410, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 432, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":408 + /* "average_inner.pyx":430 * ) * * eff_sentences, eff_words = populate_ft_s2v_config( # <<<<<<<<<<<<<< * &ft, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 408, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 430, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -4271,7 +4271,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 408, __pyx_L1_error) + __PYX_ERR(0, 430, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -4284,15 +4284,15 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_5); __Pyx_INCREF(__pyx_t_4); #else - __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 408, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 430, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); - __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 408, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 430, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 408, __pyx_L1_error) + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 430, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; @@ -4300,7 +4300,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_5); index = 1; __pyx_t_4 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_4)) goto __pyx_L9_unpacking_failed; __Pyx_GOTREF(__pyx_t_4); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 408, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 430, __pyx_L1_error) __pyx_t_7 = NULL; __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; goto __pyx_L10_unpacking_done; @@ -4308,17 +4308,17 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_t_7 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 408, __pyx_L1_error) + __PYX_ERR(0, 430, __pyx_L1_error) __pyx_L10_unpacking_done:; } - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 408, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 430, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 408, __pyx_L1_error) + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 430, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; __pyx_v_eff_sentences = __pyx_t_9; __pyx_v_eff_words = __pyx_t_8; - /* "average_inner.pyx":414 + /* "average_inner.pyx":436 * ) * * with nogil: # <<<<<<<<<<<<<< @@ -4333,7 +4333,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "average_inner.pyx":415 + /* "average_inner.pyx":437 * * with nogil: * compute_ft_sentence_averages( # <<<<<<<<<<<<<< @@ -4343,7 +4343,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_f_13average_inner_compute_ft_sentence_averages((&__pyx_v_ft), __pyx_v_eff_sentences); } - /* "average_inner.pyx":414 + /* "average_inner.pyx":436 * ) * * with nogil: # <<<<<<<<<<<<<< @@ -4364,7 +4364,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec } __pyx_L3:; - /* "average_inner.pyx":420 + /* "average_inner.pyx":442 * ) * * return eff_sentences, eff_words # <<<<<<<<<<<<<< @@ -4372,11 +4372,11 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec * def init(): */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 420, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 442, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 420, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 442, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 420, __pyx_L1_error) + __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 442, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_1); @@ -4388,8 +4388,8 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_t_5 = 0; goto __pyx_L0; - /* "average_inner.pyx":352 - * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + /* "average_inner.pyx":374 + * ) * * def train_average_cy( # <<<<<<<<<<<<<< * model, @@ -4410,7 +4410,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec return __pyx_r; } -/* "average_inner.pyx":422 +/* "average_inner.pyx":444 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -4438,7 +4438,7 @@ static PyObject *__pyx_pf_13average_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("init", 0); - /* "average_inner.pyx":423 + /* "average_inner.pyx":445 * * def init(): * return 1 # <<<<<<<<<<<<<< @@ -4450,7 +4450,7 @@ static PyObject *__pyx_pf_13average_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __pyx_r = __pyx_int_1; goto __pyx_L0; - /* "average_inner.pyx":422 + /* "average_inner.pyx":444 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -6997,8 +6997,8 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = { {0, 0, 0, 0, 0, 0, 0} }; static CYTHON_SMALL_CODE int __Pyx_InitCachedBuiltins(void) { - __pyx_builtin_enumerate = __Pyx_GetBuiltinName(__pyx_n_s_enumerate); if (!__pyx_builtin_enumerate) __PYX_ERR(0, 216, __pyx_L1_error) - __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 259, __pyx_L1_error) + __pyx_builtin_enumerate = __Pyx_GetBuiltinName(__pyx_n_s_enumerate); if (!__pyx_builtin_enumerate) __PYX_ERR(0, 217, __pyx_L1_error) + __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 260, __pyx_L1_error) __pyx_builtin_ValueError = __Pyx_GetBuiltinName(__pyx_n_s_ValueError); if (!__pyx_builtin_ValueError) __PYX_ERR(1, 272, __pyx_L1_error) __pyx_builtin_RuntimeError = __Pyx_GetBuiltinName(__pyx_n_s_RuntimeError); if (!__pyx_builtin_RuntimeError) __PYX_ERR(1, 856, __pyx_L1_error) __pyx_builtin_ImportError = __Pyx_GetBuiltinName(__pyx_n_s_ImportError); if (!__pyx_builtin_ImportError) __PYX_ERR(1, 1038, __pyx_L1_error) @@ -7088,26 +7088,26 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { __Pyx_GOTREF(__pyx_tuple__7); __Pyx_GIVEREF(__pyx_tuple__7); - /* "average_inner.pyx":352 - * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + /* "average_inner.pyx":374 + * ) * * def train_average_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 352, __pyx_L1_error) + __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 374, __pyx_L1_error) __Pyx_GOTREF(__pyx_tuple__9); __Pyx_GIVEREF(__pyx_tuple__9); - __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_train_average_cy, 352, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 352, __pyx_L1_error) + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_train_average_cy, 374, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 374, __pyx_L1_error) - /* "average_inner.pyx":422 + /* "average_inner.pyx":444 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_init, 422, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 422, __pyx_L1_error) + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_init, 444, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 444, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -7473,61 +7473,61 @@ if (!__Pyx_RefNanny) { __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "average_inner.pyx":23 - * from libc.string cimport memset + /* "average_inner.pyx":24 + * from libc.stdio cimport printf * * import scipy.linalg.blas as fblas # <<<<<<<<<<<<<< * * cdef saxpy_ptr saxpy=PyCObject_AsVoidPtr(fblas.saxpy._cpointer) # y += alpha * x */ - __pyx_t_2 = PyList_New(1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 23, __pyx_L1_error) + __pyx_t_2 = PyList_New(1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 24, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_INCREF(__pyx_n_s__8); __Pyx_GIVEREF(__pyx_n_s__8); PyList_SET_ITEM(__pyx_t_2, 0, __pyx_n_s__8); - __pyx_t_1 = __Pyx_Import(__pyx_n_s_scipy_linalg_blas, __pyx_t_2, -1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 23, __pyx_L1_error) + __pyx_t_1 = __Pyx_Import(__pyx_n_s_scipy_linalg_blas, __pyx_t_2, -1); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 24, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_1) < 0) __PYX_ERR(0, 23, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_1) < 0) __PYX_ERR(0, 24, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":25 + /* "average_inner.pyx":26 * import scipy.linalg.blas as fblas * * cdef saxpy_ptr saxpy=PyCObject_AsVoidPtr(fblas.saxpy._cpointer) # y += alpha * x # <<<<<<<<<<<<<< * cdef sscal_ptr sscal=PyCObject_AsVoidPtr(fblas.sscal._cpointer) # x = alpha * x * */ - __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_fblas); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 25, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_fblas); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 26, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_saxpy); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 25, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_saxpy); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 26, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_2, __pyx_n_s_cpointer); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 25, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_2, __pyx_n_s_cpointer); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 26, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; __pyx_v_13average_inner_saxpy = ((__pyx_t_13average_inner_saxpy_ptr)PyCObject_AsVoidPtr(__pyx_t_1)); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":26 + /* "average_inner.pyx":27 * * cdef saxpy_ptr saxpy=PyCObject_AsVoidPtr(fblas.saxpy._cpointer) # y += alpha * x * cdef sscal_ptr sscal=PyCObject_AsVoidPtr(fblas.sscal._cpointer) # x = alpha * x # <<<<<<<<<<<<<< * * cdef int ONE = 1 */ - __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_fblas); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 26, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_fblas); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 27, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_sscal); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 26, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_sscal); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 27, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_2, __pyx_n_s_cpointer); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 26, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_t_2, __pyx_n_s_cpointer); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 27, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; __pyx_v_13average_inner_sscal = ((__pyx_t_13average_inner_sscal_ptr)PyCObject_AsVoidPtr(__pyx_t_1)); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":28 + /* "average_inner.pyx":29 * cdef sscal_ptr sscal=PyCObject_AsVoidPtr(fblas.sscal._cpointer) # x = alpha * x * * cdef int ONE = 1 # <<<<<<<<<<<<<< @@ -7536,7 +7536,7 @@ if (!__Pyx_RefNanny) { */ __pyx_v_13average_inner_ONE = ((int)1); - /* "average_inner.pyx":29 + /* "average_inner.pyx":30 * * cdef int ONE = 1 * cdef int ZERO = 0 # <<<<<<<<<<<<<< @@ -7545,7 +7545,7 @@ if (!__Pyx_RefNanny) { */ __pyx_v_13average_inner_ZERO = ((int)0); - /* "average_inner.pyx":31 + /* "average_inner.pyx":32 * cdef int ZERO = 0 * * cdef REAL_t ONEF = 1.0 # <<<<<<<<<<<<<< @@ -7554,7 +7554,7 @@ if (!__Pyx_RefNanny) { */ __pyx_v_13average_inner_ONEF = ((__pyx_t_13average_inner_REAL_t)1.0); - /* "average_inner.pyx":32 + /* "average_inner.pyx":33 * * cdef REAL_t ONEF = 1.0 * cdef REAL_t ZEROF = 0.0 # <<<<<<<<<<<<<< @@ -7563,58 +7563,58 @@ if (!__Pyx_RefNanny) { */ __pyx_v_13average_inner_ZEROF = ((__pyx_t_13average_inner_REAL_t)0.0); - /* "average_inner.pyx":352 - * saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + /* "average_inner.pyx":374 + * ) * * def train_average_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_1train_average_cy, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 352, __pyx_L1_error) + __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_1train_average_cy, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 374, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_average_cy, __pyx_t_1) < 0) __PYX_ERR(0, 352, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_average_cy, __pyx_t_1) < 0) __PYX_ERR(0, 374, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":422 + /* "average_inner.pyx":444 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_3init, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 422, __pyx_L1_error) + __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_3init, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 444, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_1) < 0) __PYX_ERR(0, 422, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_1) < 0) __PYX_ERR(0, 444, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":425 + /* "average_inner.pyx":447 * return 1 * * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 425, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 447, __pyx_L1_error) - /* "average_inner.pyx":426 + /* "average_inner.pyx":448 * * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 426, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 448, __pyx_L1_error) - /* "average_inner.pyx":427 + /* "average_inner.pyx":449 * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() # <<<<<<<<<<<<<< */ - __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_init); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 427, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_init); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 449, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_CallNoArg(__pyx_t_1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 427, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_CallNoArg(__pyx_t_1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 449, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_2) < 0) __PYX_ERR(0, 427, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_2) < 0) __PYX_ERR(0, 449, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; /* "average_inner.pyx":1 diff --git a/fse/models/average_inner.pyx b/fse/models/average_inner.pyx index 5ceb5e0..de57ecc 100644 --- a/fse/models/average_inner.pyx +++ b/fse/models/average_inner.pyx @@ -253,7 +253,7 @@ cdef void compute_base_sentence_averages( uINT_t sent_idx, sent_start, sent_end, sent_row - uINT_t i, word_idx, word_row + uINT_t sent_pos, word_idx, word_row REAL_t sent_len, inv_count @@ -264,11 +264,11 @@ cdef void compute_base_sentence_averages( sent_end = c.sentence_boundary[sent_idx + 1] sent_len = ZEROF - for i in range(sent_start, sent_end): + for sent_pos in range(sent_start, sent_end): sent_len += ONEF - sent_row = c.sent_adresses[i] * size - word_row = c.word_indices[i] * size - word_idx = c.word_indices[i] + sent_row = c.sent_adresses[sent_pos] * size + word_row = c.word_indices[sent_pos] * size + word_idx = c.word_indices[sent_pos] saxpy( &size, @@ -317,7 +317,7 @@ cdef void compute_ft_sentence_averages( uINT_t ngram_row, ngrams - uINT_t i, j, word_idx, word_row + uINT_t sent_pos, ngram_pos, word_idx, word_row REAL_t sent_len REAL_t inv_count, inv_ngram @@ -330,25 +330,46 @@ cdef void compute_ft_sentence_averages( sent_end = c.sentence_boundary[sent_idx + 1] sent_len = ZEROF - for i in range(sent_start, sent_end): + for sent_pos in range(sent_start, sent_end): sent_len += ONEF - sent_row = c.sent_adresses[i] * size + sent_row = c.sent_adresses[sent_pos] * size - word_idx = c.word_indices[i] - ngrams = c.subwords_idx_len[i] + word_idx = c.word_indices[sent_pos] + ngrams = c.subwords_idx_len[sent_pos] if ngrams == 0: - word_row = c.word_indices[i] * size - saxpy(&size, &c.word_weights[word_idx], &c.word_vectors[word_row], &ONE, c.mem, &ONE) + word_row = c.word_indices[sent_pos] * size + saxpy( + &size, + &c.word_weights[word_idx], + &c.word_vectors[word_row], + &ONE, + c.mem, + &ONE + ) else: inv_ngram = (ONEF / ngrams) * c.oov_weight - for j in range(ngrams): - ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size - saxpy(&size, &inv_ngram, &c.ngram_vectors[ngram_row], &ONE, c.mem, &ONE) + for ngram_pos in range(ngrams): + ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size + saxpy( + &size, + &inv_ngram, + &c.ngram_vectors[ngram_row], + &ONE, + c.mem, + &ONE + ) if sent_len > ZEROF: inv_count = ONEF / sent_len - saxpy(&size, &inv_count, c.mem, &ONE, &c.sentence_vectors[sent_row], &ONE) + saxpy( + &size, + &inv_count, + c.mem, + &ONE, + &c.sentence_vectors[sent_row], + &ONE + ) def train_average_cy( model, From e191b8d6927c50f3729ca887d1d51966adab170c Mon Sep 17 00:00:00 2001 From: OliverB Date: Mon, 2 Mar 2020 18:18:14 +0100 Subject: [PATCH 32/63] Comments --- fse/models/base_s2v.py | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index 177b31a..145bcea 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -136,9 +136,15 @@ def __init__( [ ] modifiable sv_mapfile_path? [ ] models: - [ ] check feasibility first - [X] max-pooling -> easy - [X] hierarchical pooling -> easy + [ ] Check feasibility first + + [ ] Average: + [ ] Hierarchical Average + [ ] Stride Parameter: https://adeshpande3.github.io/A-Beginner%27s-Guide-To-Understanding-Convolutional-Neural-Networks-Part-2/ + [X] Max Pooling + [X] Hierarchical Max Pooling + [ ] Stride Paramter: https://adeshpande3.github.io/A-Beginner%27s-Guide-To-Understanding-Convolutional-Neural-Networks-Part-2/ + [ ] discrete cosine transform -> somewhat easy, questionable [ ] valve -> unclear, not cited enough [ ] power-means embedding -> very large dimensionalty From 14bab9da32ddd904c9f021572fa641be4de88c2f Mon Sep 17 00:00:00 2001 From: OliverB Date: Mon, 2 Mar 2020 18:18:42 +0100 Subject: [PATCH 33/63] Working np/cy w2v/ft pooling & unittests --- fse/models/pooling_inner.c | 832 +++++++++++++++++++++++++++-------- fse/models/pooling_inner.pyx | 164 ++++++- fse/test/test_pooling.py | 78 +++- 3 files changed, 847 insertions(+), 227 deletions(-) diff --git a/fse/models/pooling_inner.c b/fse/models/pooling_inner.c index a5c26a2..c6373cf 100644 --- a/fse/models/pooling_inner.c +++ b/fse/models/pooling_inner.c @@ -1727,6 +1727,7 @@ static void __pyx_f_13pooling_inner_swrmax_pool(int const *, float const *, floa static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_uINT_t); /*proto*/ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ +static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_uINT_t); /*proto*/ #define __Pyx_MODULE_NAME "pooling_inner" extern int __pyx_module_is_main_pooling_inner; int __pyx_module_is_main_pooling_inner = 0; @@ -1917,7 +1918,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; __pyx_t_13average_inner_uINT_t __pyx_v_sent_end; __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; - __pyx_t_13average_inner_uINT_t __pyx_v_i; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_pos; __pyx_t_13average_inner_uINT_t __pyx_v_word_idx; __pyx_t_13average_inner_uINT_t __pyx_v_word_row; __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; @@ -1974,60 +1975,60 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< * - * for i in range(sent_start, sent_end): + * for sent_pos in range(sent_start, sent_end): */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; /* "pooling_inner.pyx":101 * sent_len = ZEROF * - * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< + * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< * sent_len += ONEF - * sent_row = c.sent_adresses[i] * size + * sent_row = c.sent_adresses[sent_pos] * size */ __pyx_t_5 = __pyx_v_sent_end; __pyx_t_6 = __pyx_t_5; for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { - __pyx_v_i = __pyx_t_7; + __pyx_v_sent_pos = __pyx_t_7; /* "pooling_inner.pyx":102 * - * for i in range(sent_start, sent_end): + * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< - * sent_row = c.sent_adresses[i] * size - * word_row = c.word_indices[i] * size + * sent_row = c.sent_adresses[sent_pos] * size + * word_row = c.word_indices[sent_pos] * size */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); /* "pooling_inner.pyx":103 - * for i in range(sent_start, sent_end): + * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF - * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< - * word_row = c.word_indices[i] * size - * word_idx = c.word_indices[i] + * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< + * word_row = c.word_indices[sent_pos] * size + * word_idx = c.word_indices[sent_pos] */ - __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); + __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); /* "pooling_inner.pyx":104 * sent_len += ONEF - * sent_row = c.sent_adresses[i] * size - * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< - * word_idx = c.word_indices[i] + * sent_row = c.sent_adresses[sent_pos] * size + * word_row = c.word_indices[sent_pos] * size # <<<<<<<<<<<<<< + * word_idx = c.word_indices[sent_pos] * */ - __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); + __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_sent_pos]) * __pyx_v_size); /* "pooling_inner.pyx":105 - * sent_row = c.sent_adresses[i] * size - * word_row = c.word_indices[i] * size - * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< + * sent_row = c.sent_adresses[sent_pos] * size + * word_row = c.word_indices[sent_pos] * size + * word_idx = c.word_indices[sent_pos] # <<<<<<<<<<<<<< * * swrmax_pool( */ - __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); + __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_sent_pos]); /* "pooling_inner.pyx":107 - * word_idx = c.word_indices[i] + * word_idx = c.word_indices[sent_pos] * * swrmax_pool( # <<<<<<<<<<<<<< * &size, @@ -2063,8 +2064,8 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ __pyx_t_13average_inner_uINT_t __pyx_v_sent_end; __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; __pyx_t_13average_inner_uINT_t __pyx_v_window_end; - __pyx_t_13average_inner_uINT_t __pyx_v_i; - __pyx_t_13average_inner_uINT_t __pyx_v_j; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_pos; + __pyx_t_13average_inner_uINT_t __pyx_v_window_pos; __pyx_t_13average_inner_uINT_t __pyx_v_word_idx; __pyx_t_13average_inner_uINT_t __pyx_v_word_row; __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; @@ -2127,63 +2128,63 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< * - * for i in range(sent_start, sent_end): + * for sent_pos in range(sent_start, sent_end): */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; /* "pooling_inner.pyx":152 * sent_len = ZEROF * - * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< + * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< * sent_len += ONEF - * sent_row = c.sent_adresses[i] * size + * sent_row = c.sent_adresses[sent_pos] * size */ __pyx_t_5 = __pyx_v_sent_end; __pyx_t_6 = __pyx_t_5; for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { - __pyx_v_i = __pyx_t_7; + __pyx_v_sent_pos = __pyx_t_7; /* "pooling_inner.pyx":153 * - * for i in range(sent_start, sent_end): + * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< - * sent_row = c.sent_adresses[i] * size + * sent_row = c.sent_adresses[sent_pos] * size * */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); /* "pooling_inner.pyx":154 - * for i in range(sent_start, sent_end): + * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF - * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< + * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< * - * if i + window_size > sent_end: + * if sent_pos + window_size > sent_end: */ - __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); + __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); /* "pooling_inner.pyx":156 - * sent_row = c.sent_adresses[i] * size + * sent_row = c.sent_adresses[sent_pos] * size * - * if i + window_size > sent_end: # <<<<<<<<<<<<<< + * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< * window_end = sent_end * else: */ - __pyx_t_8 = (((__pyx_v_i + __pyx_v_window_size) > __pyx_v_sent_end) != 0); + __pyx_t_8 = (((__pyx_v_sent_pos + __pyx_v_window_size) > __pyx_v_sent_end) != 0); if (__pyx_t_8) { /* "pooling_inner.pyx":157 * - * if i + window_size > sent_end: + * if sent_pos + window_size > sent_end: * window_end = sent_end # <<<<<<<<<<<<<< * else: - * window_end = i + window_size + * window_end = sent_pos + window_size */ __pyx_v_window_end = __pyx_v_sent_end; /* "pooling_inner.pyx":156 - * sent_row = c.sent_adresses[i] * size + * sent_row = c.sent_adresses[sent_pos] * size * - * if i + window_size > sent_end: # <<<<<<<<<<<<<< + * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< * window_end = sent_end * else: */ @@ -2193,12 +2194,12 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ /* "pooling_inner.pyx":159 * window_end = sent_end * else: - * window_end = i + window_size # <<<<<<<<<<<<<< + * window_end = sent_pos + window_size # <<<<<<<<<<<<<< * * # Compute the locally averaged window */ /*else*/ { - __pyx_v_window_end = (__pyx_v_i + __pyx_v_window_size); + __pyx_v_window_end = (__pyx_v_sent_pos + __pyx_v_window_size); } __pyx_L7:; @@ -2216,7 +2217,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) - * for j in range(i, window_end): + * for window_pos in range(sent_pos, window_end): */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); @@ -2224,7 +2225,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< - * for j in range(i, window_end): + * for window_pos in range(sent_pos, window_end): * win_len += ONEF */ (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); @@ -2232,44 +2233,44 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ /* "pooling_inner.pyx":165 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) - * for j in range(i, window_end): # <<<<<<<<<<<<<< + * for window_pos in range(sent_pos, window_end): # <<<<<<<<<<<<<< * win_len += ONEF * */ __pyx_t_9 = __pyx_v_window_end; __pyx_t_10 = __pyx_t_9; - for (__pyx_t_11 = __pyx_v_i; __pyx_t_11 < __pyx_t_10; __pyx_t_11+=1) { - __pyx_v_j = __pyx_t_11; + for (__pyx_t_11 = __pyx_v_sent_pos; __pyx_t_11 < __pyx_t_10; __pyx_t_11+=1) { + __pyx_v_window_pos = __pyx_t_11; /* "pooling_inner.pyx":166 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) - * for j in range(i, window_end): + * for window_pos in range(sent_pos, window_end): * win_len += ONEF # <<<<<<<<<<<<<< * - * word_row = c.word_indices[j] * size + * word_row = c.word_indices[window_pos] * size */ __pyx_v_win_len = (__pyx_v_win_len + __pyx_v_13average_inner_ONEF); /* "pooling_inner.pyx":168 * win_len += ONEF * - * word_row = c.word_indices[j] * size # <<<<<<<<<<<<<< - * word_idx = c.word_indices[j] + * word_row = c.word_indices[window_pos] * size # <<<<<<<<<<<<<< + * word_idx = c.word_indices[window_pos] * */ - __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_j]) * __pyx_v_size); + __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_window_pos]) * __pyx_v_size); /* "pooling_inner.pyx":169 * - * word_row = c.word_indices[j] * size - * word_idx = c.word_indices[j] # <<<<<<<<<<<<<< + * word_row = c.word_indices[window_pos] * size + * word_idx = c.word_indices[window_pos] # <<<<<<<<<<<<<< * * saxpy( */ - __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_j]); + __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_window_pos]); /* "pooling_inner.pyx":171 - * word_idx = c.word_indices[j] + * word_idx = c.word_indices[window_pos] * * saxpy( # <<<<<<<<<<<<<< * &size, @@ -2353,8 +2354,8 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; __pyx_t_13average_inner_uINT_t __pyx_v_ngram_row; __pyx_t_13average_inner_uINT_t __pyx_v_ngrams; - __pyx_t_13average_inner_uINT_t __pyx_v_i; - __pyx_t_13average_inner_uINT_t __pyx_v_j; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_pos; + __pyx_t_13average_inner_uINT_t __pyx_v_ngram_pos; __pyx_t_13average_inner_uINT_t __pyx_v_word_idx; __pyx_t_13average_inner_uINT_t __pyx_v_word_row; __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; @@ -2428,63 +2429,63 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< * - * for i in range(sent_start, sent_end): + * for sent_pos in range(sent_start, sent_end): */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; /* "pooling_inner.pyx":235 * sent_len = ZEROF * - * for i in range(sent_start, sent_end): # <<<<<<<<<<<<<< + * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< * sent_len += ONEF - * sent_row = c.sent_adresses[i] * size + * sent_row = c.sent_adresses[sent_pos] * size */ __pyx_t_6 = __pyx_v_sent_end; __pyx_t_7 = __pyx_t_6; for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { - __pyx_v_i = __pyx_t_8; + __pyx_v_sent_pos = __pyx_t_8; /* "pooling_inner.pyx":236 * - * for i in range(sent_start, sent_end): + * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< - * sent_row = c.sent_adresses[i] * size + * sent_row = c.sent_adresses[sent_pos] * size * */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); /* "pooling_inner.pyx":237 - * for i in range(sent_start, sent_end): + * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF - * sent_row = c.sent_adresses[i] * size # <<<<<<<<<<<<<< + * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< * - * word_idx = c.word_indices[i] + * word_idx = c.word_indices[sent_pos] */ - __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_i]) * __pyx_v_size); + __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); /* "pooling_inner.pyx":239 - * sent_row = c.sent_adresses[i] * size + * sent_row = c.sent_adresses[sent_pos] * size * - * word_idx = c.word_indices[i] # <<<<<<<<<<<<<< - * ngrams = c.subwords_idx_len[i] + * word_idx = c.word_indices[sent_pos] # <<<<<<<<<<<<<< + * ngrams = c.subwords_idx_len[sent_pos] * */ - __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_i]); + __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_sent_pos]); /* "pooling_inner.pyx":240 * - * word_idx = c.word_indices[i] - * ngrams = c.subwords_idx_len[i] # <<<<<<<<<<<<<< + * word_idx = c.word_indices[sent_pos] + * ngrams = c.subwords_idx_len[sent_pos] # <<<<<<<<<<<<<< * * if ngrams == 0: */ - __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_i]); + __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_sent_pos]); /* "pooling_inner.pyx":242 - * ngrams = c.subwords_idx_len[i] + * ngrams = c.subwords_idx_len[sent_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< - * word_row = c.word_indices[i] * size + * word_row = c.word_indices[sent_pos] * size * */ __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); @@ -2493,14 +2494,14 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 /* "pooling_inner.pyx":243 * * if ngrams == 0: - * word_row = c.word_indices[i] * size # <<<<<<<<<<<<<< + * word_row = c.word_indices[sent_pos] * size # <<<<<<<<<<<<<< * * swrmax_pool( */ - __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_i]) * __pyx_v_size); + __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_sent_pos]) * __pyx_v_size); /* "pooling_inner.pyx":245 - * word_row = c.word_indices[i] * size + * word_row = c.word_indices[sent_pos] * size * * swrmax_pool( # <<<<<<<<<<<<<< * &size, @@ -2509,10 +2510,10 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_f_13pooling_inner_swrmax_pool((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row]))); /* "pooling_inner.pyx":242 - * ngrams = c.subwords_idx_len[i] + * ngrams = c.subwords_idx_len[sent_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< - * word_row = c.word_indices[i] * size + * word_row = c.word_indices[sent_pos] * size * */ goto __pyx_L7; @@ -2523,7 +2524,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 * else: * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< * inv_ngram = (ONEF / ngrams) * c.oov_weight - * for j in range(ngrams): + * for ngram_pos in range(ngrams): */ /*else*/ { (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); @@ -2532,35 +2533,35 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 * else: * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< - * for j in range(ngrams): - * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size + * for ngram_pos in range(ngrams): + * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size */ __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); /* "pooling_inner.pyx":255 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight - * for j in range(ngrams): # <<<<<<<<<<<<<< - * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size + * for ngram_pos in range(ngrams): # <<<<<<<<<<<<<< + * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size * saxpy( */ __pyx_t_10 = __pyx_v_ngrams; __pyx_t_11 = __pyx_t_10; for (__pyx_t_12 = 0; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { - __pyx_v_j = __pyx_t_12; + __pyx_v_ngram_pos = __pyx_t_12; /* "pooling_inner.pyx":256 * inv_ngram = (ONEF / ngrams) * c.oov_weight - * for j in range(ngrams): - * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size # <<<<<<<<<<<<<< + * for ngram_pos in range(ngrams): + * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size # <<<<<<<<<<<<<< * saxpy( * &size, */ - __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_i * 40) + __pyx_v_j)]) * __pyx_v_size); + __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_sent_pos * 40) + __pyx_v_ngram_pos)]) * __pyx_v_size); /* "pooling_inner.pyx":257 - * for j in range(ngrams): - * ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size + * for ngram_pos in range(ngrams): + * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size * saxpy( # <<<<<<<<<<<<<< * &size, * &inv_ngram, @@ -2592,7 +2593,414 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 /* function exit code */ } -/* "pooling_inner.pyx":274 +/* "pooling_inner.pyx":275 + * + * + * cdef void compute_ft_sentence_hier_pooling( # <<<<<<<<<<<<<< + * FTSentenceVecsConfig *c, + * uINT_t num_sentences, + */ + +static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences, __pyx_t_13average_inner_uINT_t __pyx_v_window_size) { + int __pyx_v_size; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_end; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_row; + __pyx_t_13average_inner_uINT_t __pyx_v_ngram_row; + __pyx_t_13average_inner_uINT_t __pyx_v_ngrams; + __pyx_t_13average_inner_uINT_t __pyx_v_sent_pos; + __pyx_t_13average_inner_uINT_t __pyx_v_window_pos; + __pyx_t_13average_inner_uINT_t __pyx_v_ngram_pos; + __pyx_t_13average_inner_uINT_t __pyx_v_word_idx; + __pyx_t_13average_inner_uINT_t __pyx_v_word_row; + __pyx_t_13average_inner_REAL_t __pyx_v_sent_len; + __pyx_t_13average_inner_REAL_t __pyx_v_win_len; + __pyx_t_13average_inner_REAL_t __pyx_v_inv_count; + __pyx_t_13average_inner_REAL_t __pyx_v_inv_ngram; + __pyx_t_13average_inner_REAL_t __pyx_v_oov_weight; + __pyx_t_13average_inner_uINT_t __pyx_v_window_end; + int __pyx_t_1; + __pyx_t_13average_inner_REAL_t __pyx_t_2; + __pyx_t_13average_inner_uINT_t __pyx_t_3; + __pyx_t_13average_inner_uINT_t __pyx_t_4; + __pyx_t_13average_inner_uINT_t __pyx_t_5; + __pyx_t_13average_inner_uINT_t __pyx_t_6; + __pyx_t_13average_inner_uINT_t __pyx_t_7; + __pyx_t_13average_inner_uINT_t __pyx_t_8; + int __pyx_t_9; + __pyx_t_13average_inner_uINT_t __pyx_t_10; + __pyx_t_13average_inner_uINT_t __pyx_t_11; + __pyx_t_13average_inner_uINT_t __pyx_t_12; + __pyx_t_13average_inner_uINT_t __pyx_t_13; + __pyx_t_13average_inner_uINT_t __pyx_t_14; + __pyx_t_13average_inner_uINT_t __pyx_t_15; + + /* "pooling_inner.pyx":299 + * + * cdef: + * int size = c.size # <<<<<<<<<<<<<< + * + * uINT_t sent_idx, sent_start, sent_end, sent_row + */ + __pyx_t_1 = __pyx_v_c->size; + __pyx_v_size = __pyx_t_1; + + /* "pooling_inner.pyx":309 + * REAL_t sent_len, win_len + * REAL_t inv_count, inv_ngram + * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< + * + * for sent_idx in range(num_sentences): + */ + __pyx_t_2 = __pyx_v_c->oov_weight; + __pyx_v_oov_weight = __pyx_t_2; + + /* "pooling_inner.pyx":311 + * REAL_t oov_weight = c.oov_weight + * + * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] + */ + __pyx_t_3 = __pyx_v_num_sentences; + __pyx_t_4 = __pyx_t_3; + for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { + __pyx_v_sent_idx = __pyx_t_5; + + /* "pooling_inner.pyx":312 + * + * for sent_idx in range(num_sentences): + * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< + * sent_end = c.sentence_boundary[sent_idx + 1] + * sent_len = ZEROF + */ + __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); + + /* "pooling_inner.pyx":313 + * for sent_idx in range(num_sentences): + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< + * sent_len = ZEROF + * + */ + __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); + + /* "pooling_inner.pyx":314 + * sent_start = c.sentence_boundary[sent_idx] + * sent_end = c.sentence_boundary[sent_idx + 1] + * sent_len = ZEROF # <<<<<<<<<<<<<< + * + * for sent_pos in range(sent_start, sent_end): + */ + __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; + + /* "pooling_inner.pyx":316 + * sent_len = ZEROF + * + * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< + * sent_len += ONEF + * sent_row = c.sent_adresses[sent_pos] * size + */ + __pyx_t_6 = __pyx_v_sent_end; + __pyx_t_7 = __pyx_t_6; + for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { + __pyx_v_sent_pos = __pyx_t_8; + + /* "pooling_inner.pyx":317 + * + * for sent_pos in range(sent_start, sent_end): + * sent_len += ONEF # <<<<<<<<<<<<<< + * sent_row = c.sent_adresses[sent_pos] * size + * + */ + __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); + + /* "pooling_inner.pyx":318 + * for sent_pos in range(sent_start, sent_end): + * sent_len += ONEF + * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< + * + * if sent_pos + window_size > sent_end: + */ + __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); + + /* "pooling_inner.pyx":320 + * sent_row = c.sent_adresses[sent_pos] * size + * + * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< + * window_end = sent_end + * else: + */ + __pyx_t_9 = (((__pyx_v_sent_pos + __pyx_v_window_size) > __pyx_v_sent_end) != 0); + if (__pyx_t_9) { + + /* "pooling_inner.pyx":321 + * + * if sent_pos + window_size > sent_end: + * window_end = sent_end # <<<<<<<<<<<<<< + * else: + * window_end = sent_pos + window_size + */ + __pyx_v_window_end = __pyx_v_sent_end; + + /* "pooling_inner.pyx":320 + * sent_row = c.sent_adresses[sent_pos] * size + * + * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< + * window_end = sent_end + * else: + */ + goto __pyx_L7; + } + + /* "pooling_inner.pyx":323 + * window_end = sent_end + * else: + * window_end = sent_pos + window_size # <<<<<<<<<<<<<< + * + * # Compute the locally averaged window + */ + /*else*/ { + __pyx_v_window_end = (__pyx_v_sent_pos + __pyx_v_window_size); + } + __pyx_L7:; + + /* "pooling_inner.pyx":326 + * + * # Compute the locally averaged window + * win_len = ZEROF # <<<<<<<<<<<<<< + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + */ + __pyx_v_win_len = __pyx_v_13average_inner_ZEROF; + + /* "pooling_inner.pyx":327 + * # Compute the locally averaged window + * win_len = ZEROF + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + * for window_pos in range(sent_pos, window_end): + */ + (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); + + /* "pooling_inner.pyx":328 + * win_len = ZEROF + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< + * for window_pos in range(sent_pos, window_end): + * win_len += ONEF + */ + (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); + + /* "pooling_inner.pyx":329 + * memset(c.mem, 0, size * cython.sizeof(REAL_t)) + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + * for window_pos in range(sent_pos, window_end): # <<<<<<<<<<<<<< + * win_len += ONEF + * ngrams = c.subwords_idx_len[window_pos] + */ + __pyx_t_10 = __pyx_v_window_end; + __pyx_t_11 = __pyx_t_10; + for (__pyx_t_12 = __pyx_v_sent_pos; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { + __pyx_v_window_pos = __pyx_t_12; + + /* "pooling_inner.pyx":330 + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + * for window_pos in range(sent_pos, window_end): + * win_len += ONEF # <<<<<<<<<<<<<< + * ngrams = c.subwords_idx_len[window_pos] + * + */ + __pyx_v_win_len = (__pyx_v_win_len + __pyx_v_13average_inner_ONEF); + + /* "pooling_inner.pyx":331 + * for window_pos in range(sent_pos, window_end): + * win_len += ONEF + * ngrams = c.subwords_idx_len[window_pos] # <<<<<<<<<<<<<< + * + * if ngrams == 0: + */ + __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_window_pos]); + + /* "pooling_inner.pyx":333 + * ngrams = c.subwords_idx_len[window_pos] + * + * if ngrams == 0: # <<<<<<<<<<<<<< + * word_row = c.word_indices[window_pos] * size + * word_idx = c.word_indices[window_pos] + */ + __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); + if (__pyx_t_9) { + + /* "pooling_inner.pyx":334 + * + * if ngrams == 0: + * word_row = c.word_indices[window_pos] * size # <<<<<<<<<<<<<< + * word_idx = c.word_indices[window_pos] + * + */ + __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_window_pos]) * __pyx_v_size); + + /* "pooling_inner.pyx":335 + * if ngrams == 0: + * word_row = c.word_indices[window_pos] * size + * word_idx = c.word_indices[window_pos] # <<<<<<<<<<<<<< + * + * saxpy( + */ + __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_window_pos]); + + /* "pooling_inner.pyx":337 + * word_idx = c.word_indices[window_pos] + * + * saxpy( # <<<<<<<<<<<<<< + * &size, + * &c.word_weights[word_idx], + */ + __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); + + /* "pooling_inner.pyx":333 + * ngrams = c.subwords_idx_len[window_pos] + * + * if ngrams == 0: # <<<<<<<<<<<<<< + * word_row = c.word_indices[window_pos] * size + * word_idx = c.word_indices[window_pos] + */ + goto __pyx_L10; + } + + /* "pooling_inner.pyx":347 + * + * else: + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< + * inv_ngram = (ONEF / ngrams) * c.oov_weight + * for ngram_pos in range(ngrams): + */ + /*else*/ { + (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); + + /* "pooling_inner.pyx":348 + * else: + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< + * for ngram_pos in range(ngrams): + * ngram_row = c.subwords_idx[(window_pos * MAX_NGRAMS)+ngram_pos] * size + */ + __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); + + /* "pooling_inner.pyx":349 + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + * inv_ngram = (ONEF / ngrams) * c.oov_weight + * for ngram_pos in range(ngrams): # <<<<<<<<<<<<<< + * ngram_row = c.subwords_idx[(window_pos * MAX_NGRAMS)+ngram_pos] * size + * saxpy( + */ + __pyx_t_13 = __pyx_v_ngrams; + __pyx_t_14 = __pyx_t_13; + for (__pyx_t_15 = 0; __pyx_t_15 < __pyx_t_14; __pyx_t_15+=1) { + __pyx_v_ngram_pos = __pyx_t_15; + + /* "pooling_inner.pyx":350 + * inv_ngram = (ONEF / ngrams) * c.oov_weight + * for ngram_pos in range(ngrams): + * ngram_row = c.subwords_idx[(window_pos * MAX_NGRAMS)+ngram_pos] * size # <<<<<<<<<<<<<< + * saxpy( + * &size, + */ + __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_window_pos * 40) + __pyx_v_ngram_pos)]) * __pyx_v_size); + + /* "pooling_inner.pyx":351 + * for ngram_pos in range(ngrams): + * ngram_row = c.subwords_idx[(window_pos * MAX_NGRAMS)+ngram_pos] * size + * saxpy( # <<<<<<<<<<<<<< + * &size, + * &inv_ngram, + */ + __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_ngram), (&(__pyx_v_c->ngram_vectors[__pyx_v_ngram_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE)); + } + + /* "pooling_inner.pyx":360 + * ) + * + * saxpy( # <<<<<<<<<<<<<< + * &size, + * &oov_weight, + */ + __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_oov_weight), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); + } + __pyx_L10:; + } + + /* "pooling_inner.pyx":369 + * ) + * + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< + * # Rescale for dynamic window size + * if win_len > ZEROF: + */ + (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); + + /* "pooling_inner.pyx":371 + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + * # Rescale for dynamic window size + * if win_len > ZEROF: # <<<<<<<<<<<<<< + * inv_count = ONEF / win_len + * saxpy( + */ + __pyx_t_9 = ((__pyx_v_win_len > __pyx_v_13average_inner_ZEROF) != 0); + if (__pyx_t_9) { + + /* "pooling_inner.pyx":372 + * # Rescale for dynamic window size + * if win_len > ZEROF: + * inv_count = ONEF / win_len # <<<<<<<<<<<<<< + * saxpy( + * &size, + */ + __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_win_len); + + /* "pooling_inner.pyx":373 + * if win_len > ZEROF: + * inv_count = ONEF / win_len + * saxpy( # <<<<<<<<<<<<<< + * &size, + * &inv_count, + */ + __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE)); + + /* "pooling_inner.pyx":371 + * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + * # Rescale for dynamic window size + * if win_len > ZEROF: # <<<<<<<<<<<<<< + * inv_count = ONEF / win_len + * saxpy( + */ + } + + /* "pooling_inner.pyx":382 + * ) + * + * swrmax_pool( # <<<<<<<<<<<<<< + * &size, + * &ONEF, + */ + __pyx_f_13pooling_inner_swrmax_pool((&__pyx_v_size), (&__pyx_v_13average_inner_ONEF), __pyx_v_c->mem2, (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row]))); + } + } + + /* "pooling_inner.pyx":275 + * + * + * cdef void compute_ft_sentence_hier_pooling( # <<<<<<<<<<<<<< + * FTSentenceVecsConfig *c, + * uINT_t num_sentences, + */ + + /* function exit code */ +} + +/* "pooling_inner.pyx":390 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< @@ -2639,23 +3047,23 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self case 1: if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 274, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 390, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 2: if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 274, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 390, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 3: if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 274, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 390, __pyx_L3_error) } } if (unlikely(kw_args > 0)) { - if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 274, __pyx_L3_error) + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 390, __pyx_L3_error) } } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { goto __pyx_L5_argtuple_error; @@ -2672,7 +3080,7 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self } goto __pyx_L4_argument_unpacking_done; __pyx_L5_argtuple_error:; - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 274, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 390, __pyx_L3_error) __pyx_L3_error:; __Pyx_AddTraceback("pooling_inner.train_pooling_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __Pyx_RefNannyFinishContext(); @@ -2704,7 +3112,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_13average_inner_uINT_t __pyx_t_9; __Pyx_RefNannySetupContext("train_pooling_cy", 0); - /* "pooling_inner.pyx":303 + /* "pooling_inner.pyx":419 * """ * * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< @@ -2713,7 +3121,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_sentences = 0; - /* "pooling_inner.pyx":304 + /* "pooling_inner.pyx":420 * * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< @@ -2722,65 +3130,65 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_words = 0; - /* "pooling_inner.pyx":305 + /* "pooling_inner.pyx":421 * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 * cdef uINT_t window_size = model.window_size # <<<<<<<<<<<<<< * cdef BaseSentenceVecsConfig w2v * cdef FTSentenceVecsConfig ft */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_size); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 305, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_size); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 421, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_1); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 305, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_1); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 421, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_v_window_size = ((__pyx_t_13average_inner_uINT_t)__pyx_t_2); - /* "pooling_inner.pyx":309 + /* "pooling_inner.pyx":425 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< * init_base_s2v_config(&w2v, model, target, memory) * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 309, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 425, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 309, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 425, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_4 = ((!__pyx_t_3) != 0); if (__pyx_t_4) { - /* "pooling_inner.pyx":310 + /* "pooling_inner.pyx":426 * * if not model.is_ft: * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< * * eff_sentences, eff_words = populate_base_s2v_config( */ - __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 310, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 426, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":314 + /* "pooling_inner.pyx":430 * eff_sentences, eff_words = populate_base_s2v_config( * &w2v, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 314, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 430, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 314, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 430, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":312 + /* "pooling_inner.pyx":428 * init_base_s2v_config(&w2v, model, target, memory) * * eff_sentences, eff_words = populate_base_s2v_config( # <<<<<<<<<<<<<< * &w2v, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 312, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 428, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -2789,7 +3197,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 312, __pyx_L1_error) + __PYX_ERR(0, 428, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -2802,15 +3210,15 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_5); __Pyx_INCREF(__pyx_t_6); #else - __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 312, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 428, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); - __pyx_t_6 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 312, __pyx_L1_error) + __pyx_t_6 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 428, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 312, __pyx_L1_error) + __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 428, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_8 = Py_TYPE(__pyx_t_7)->tp_iternext; @@ -2818,7 +3226,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_5); index = 1; __pyx_t_6 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_6)) goto __pyx_L4_unpacking_failed; __Pyx_GOTREF(__pyx_t_6); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 312, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 428, __pyx_L1_error) __pyx_t_8 = NULL; __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; goto __pyx_L5_unpacking_done; @@ -2826,31 +3234,31 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; __pyx_t_8 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 312, __pyx_L1_error) + __PYX_ERR(0, 428, __pyx_L1_error) __pyx_L5_unpacking_done:; } - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 312, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 428, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 312, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 428, __pyx_L1_error) __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_v_eff_sentences = __pyx_t_2; __pyx_v_eff_words = __pyx_t_9; - /* "pooling_inner.pyx":318 + /* "pooling_inner.pyx":434 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: * compute_base_sentence_pooling( */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 318, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 434, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 318, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 434, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_3 = ((!__pyx_t_4) != 0); if (__pyx_t_3) { - /* "pooling_inner.pyx":319 + /* "pooling_inner.pyx":435 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -2865,7 +3273,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":320 + /* "pooling_inner.pyx":436 * if not model.hierarchical: * with nogil: * compute_base_sentence_pooling( # <<<<<<<<<<<<<< @@ -2875,7 +3283,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_base_sentence_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":319 + /* "pooling_inner.pyx":435 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -2894,7 +3302,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } } - /* "pooling_inner.pyx":318 + /* "pooling_inner.pyx":434 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< @@ -2904,7 +3312,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L6; } - /* "pooling_inner.pyx":325 + /* "pooling_inner.pyx":441 * ) * else: * with nogil: # <<<<<<<<<<<<<< @@ -2920,7 +3328,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":326 + /* "pooling_inner.pyx":442 * else: * with nogil: * compute_base_sentence_hier_pooling( # <<<<<<<<<<<<<< @@ -2930,7 +3338,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences, __pyx_v_window_size); } - /* "pooling_inner.pyx":325 + /* "pooling_inner.pyx":441 * ) * else: * with nogil: # <<<<<<<<<<<<<< @@ -2951,7 +3359,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } __pyx_L6:; - /* "pooling_inner.pyx":309 + /* "pooling_inner.pyx":425 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< @@ -2961,7 +3369,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L3; } - /* "pooling_inner.pyx":332 + /* "pooling_inner.pyx":448 * ) * else: * init_ft_s2v_config(&ft, model, target, memory) # <<<<<<<<<<<<<< @@ -2969,31 +3377,31 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec * eff_sentences, eff_words = populate_ft_s2v_config( */ /*else*/ { - __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 332, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 448, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":336 + /* "pooling_inner.pyx":452 * eff_sentences, eff_words = populate_ft_s2v_config( * &ft, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 336, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 452, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_6 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 336, __pyx_L1_error) + __pyx_t_6 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 452, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":334 + /* "pooling_inner.pyx":450 * init_ft_s2v_config(&ft, model, target, memory) * * eff_sentences, eff_words = populate_ft_s2v_config( # <<<<<<<<<<<<<< * &ft, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_6, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 334, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_6, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 450, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -3002,7 +3410,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 334, __pyx_L1_error) + __PYX_ERR(0, 450, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -3015,15 +3423,15 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_6); __Pyx_INCREF(__pyx_t_5); #else - __pyx_t_6 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 334, __pyx_L1_error) + __pyx_t_6 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 450, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); - __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 334, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 450, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 334, __pyx_L1_error) + __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 450, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_8 = Py_TYPE(__pyx_t_7)->tp_iternext; @@ -3031,7 +3439,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_6); index = 1; __pyx_t_5 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_5)) goto __pyx_L13_unpacking_failed; __Pyx_GOTREF(__pyx_t_5); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 334, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 450, __pyx_L1_error) __pyx_t_8 = NULL; __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; goto __pyx_L14_unpacking_done; @@ -3039,36 +3447,36 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; __pyx_t_8 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 334, __pyx_L1_error) + __PYX_ERR(0, 450, __pyx_L1_error) __pyx_L14_unpacking_done:; } - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 334, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 450, __pyx_L1_error) __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 334, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 450, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; __pyx_v_eff_sentences = __pyx_t_9; __pyx_v_eff_words = __pyx_t_2; - /* "pooling_inner.pyx":340 + /* "pooling_inner.pyx":456 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: * compute_ft_sentence_pooling(&ft, eff_sentences) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 340, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 456, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 340, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 456, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_4 = ((!__pyx_t_3) != 0); if (__pyx_t_4) { - /* "pooling_inner.pyx":341 + /* "pooling_inner.pyx":457 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< * compute_ft_sentence_pooling(&ft, eff_sentences) - * + * else: */ { #ifdef WITH_THREAD @@ -3078,22 +3486,22 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":342 + /* "pooling_inner.pyx":458 * if not model.hierarchical: * with nogil: * compute_ft_sentence_pooling(&ft, eff_sentences) # <<<<<<<<<<<<<< - * - * return eff_sentences, eff_words + * else: + * with nogil: */ __pyx_f_13pooling_inner_compute_ft_sentence_pooling((&__pyx_v_ft), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":341 + /* "pooling_inner.pyx":457 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< * compute_ft_sentence_pooling(&ft, eff_sentences) - * + * else: */ /*finally:*/ { /*normal exit:*/{ @@ -3107,30 +3515,78 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } } - /* "pooling_inner.pyx":340 + /* "pooling_inner.pyx":456 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: * compute_ft_sentence_pooling(&ft, eff_sentences) */ + goto __pyx_L15; } + + /* "pooling_inner.pyx":460 + * compute_ft_sentence_pooling(&ft, eff_sentences) + * else: + * with nogil: # <<<<<<<<<<<<<< + * compute_ft_sentence_hier_pooling( + * &ft, + */ + /*else*/ { + { + #ifdef WITH_THREAD + PyThreadState *_save; + Py_UNBLOCK_THREADS + __Pyx_FastGIL_Remember(); + #endif + /*try:*/ { + + /* "pooling_inner.pyx":461 + * else: + * with nogil: + * compute_ft_sentence_hier_pooling( # <<<<<<<<<<<<<< + * &ft, + * eff_sentences, + */ + __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling((&__pyx_v_ft), __pyx_v_eff_sentences, __pyx_v_window_size); + } + + /* "pooling_inner.pyx":460 + * compute_ft_sentence_pooling(&ft, eff_sentences) + * else: + * with nogil: # <<<<<<<<<<<<<< + * compute_ft_sentence_hier_pooling( + * &ft, + */ + /*finally:*/ { + /*normal exit:*/{ + #ifdef WITH_THREAD + __Pyx_FastGIL_Forget(); + Py_BLOCK_THREADS + #endif + goto __pyx_L21; + } + __pyx_L21:; + } + } + } + __pyx_L15:; } __pyx_L3:; - /* "pooling_inner.pyx":344 - * compute_ft_sentence_pooling(&ft, eff_sentences) + /* "pooling_inner.pyx":467 + * ) * * return eff_sentences, eff_words # <<<<<<<<<<<<<< * * def init(): */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 344, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 467, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 344, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 467, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); - __pyx_t_6 = PyTuple_New(2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 344, __pyx_L1_error) + __pyx_t_6 = PyTuple_New(2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 467, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_6, 0, __pyx_t_1); @@ -3142,7 +3598,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_6 = 0; goto __pyx_L0; - /* "pooling_inner.pyx":274 + /* "pooling_inner.pyx":390 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< @@ -3164,7 +3620,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec return __pyx_r; } -/* "pooling_inner.pyx":346 +/* "pooling_inner.pyx":469 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -3192,7 +3648,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("init", 0); - /* "pooling_inner.pyx":347 + /* "pooling_inner.pyx":470 * * def init(): * return 1 # <<<<<<<<<<<<<< @@ -3204,7 +3660,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __pyx_r = __pyx_int_1; goto __pyx_L0; - /* "pooling_inner.pyx":346 + /* "pooling_inner.pyx":469 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -5818,26 +6274,26 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { __Pyx_GOTREF(__pyx_tuple__7); __Pyx_GIVEREF(__pyx_tuple__7); - /* "pooling_inner.pyx":274 + /* "pooling_inner.pyx":390 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_tuple__9 = PyTuple_Pack(9, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_window_size, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 274, __pyx_L1_error) + __pyx_tuple__9 = PyTuple_Pack(9, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_window_size, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 390, __pyx_L1_error) __Pyx_GOTREF(__pyx_tuple__9); __Pyx_GIVEREF(__pyx_tuple__9); - __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 9, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 274, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 274, __pyx_L1_error) + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 9, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 390, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 390, __pyx_L1_error) - /* "pooling_inner.pyx":346 + /* "pooling_inner.pyx":469 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 346, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 346, __pyx_L1_error) + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 469, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 469, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -6200,58 +6656,58 @@ if (!__Pyx_RefNanny) { if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_2) < 0) __PYX_ERR(0, 22, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":274 + /* "pooling_inner.pyx":390 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 274, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 390, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 274, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 390, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":346 + /* "pooling_inner.pyx":469 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 346, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 469, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 346, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 469, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":349 + /* "pooling_inner.pyx":472 * return 1 * * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 349, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 472, __pyx_L1_error) - /* "pooling_inner.pyx":350 + /* "pooling_inner.pyx":473 * * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 350, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 473, __pyx_L1_error) - /* "pooling_inner.pyx":351 + /* "pooling_inner.pyx":474 * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() # <<<<<<<<<<<<<< */ - __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 351, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 474, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 351, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 474, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 351, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 474, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; /* "pooling_inner.pyx":1 diff --git a/fse/models/pooling_inner.pyx b/fse/models/pooling_inner.pyx index ff6fd69..30be333 100644 --- a/fse/models/pooling_inner.pyx +++ b/fse/models/pooling_inner.pyx @@ -89,7 +89,7 @@ cdef void compute_base_sentence_pooling( uINT_t sent_idx, sent_start, sent_end, sent_row - uINT_t i, word_idx, word_row + uINT_t sent_pos, word_idx, word_row REAL_t sent_len, inv_count @@ -98,11 +98,11 @@ cdef void compute_base_sentence_pooling( sent_end = c.sentence_boundary[sent_idx + 1] sent_len = ZEROF - for i in range(sent_start, sent_end): + for sent_pos in range(sent_start, sent_end): sent_len += ONEF - sent_row = c.sent_adresses[i] * size - word_row = c.word_indices[i] * size - word_idx = c.word_indices[i] + sent_row = c.sent_adresses[sent_pos] * size + word_row = c.word_indices[sent_pos] * size + word_idx = c.word_indices[sent_pos] swrmax_pool( &size, @@ -140,7 +140,7 @@ cdef void compute_base_sentence_hier_pooling( uINT_t sent_idx, sent_start, sent_end, sent_row, window_end - uINT_t i, j, word_idx, word_row + uINT_t sent_pos, window_pos, word_idx, word_row REAL_t sent_len, win_len, inv_count @@ -149,24 +149,24 @@ cdef void compute_base_sentence_hier_pooling( sent_end = c.sentence_boundary[sent_idx + 1] sent_len = ZEROF - for i in range(sent_start, sent_end): + for sent_pos in range(sent_start, sent_end): sent_len += ONEF - sent_row = c.sent_adresses[i] * size + sent_row = c.sent_adresses[sent_pos] * size - if i + window_size > sent_end: + if sent_pos + window_size > sent_end: window_end = sent_end else: - window_end = i + window_size + window_end = sent_pos + window_size # Compute the locally averaged window win_len = ZEROF memset(c.mem, 0, size * cython.sizeof(REAL_t)) memset(c.mem2, 0, size * cython.sizeof(REAL_t)) - for j in range(i, window_end): + for window_pos in range(sent_pos, window_end): win_len += ONEF - word_row = c.word_indices[j] * size - word_idx = c.word_indices[j] + word_row = c.word_indices[window_pos] * size + word_idx = c.word_indices[window_pos] saxpy( &size, @@ -221,7 +221,7 @@ cdef void compute_ft_sentence_pooling( uINT_t ngram_row, ngrams - uINT_t i, j, word_idx, word_row + uINT_t sent_pos, ngram_pos, word_idx, word_row REAL_t sent_len REAL_t inv_count, inv_ngram @@ -232,15 +232,15 @@ cdef void compute_ft_sentence_pooling( sent_end = c.sentence_boundary[sent_idx + 1] sent_len = ZEROF - for i in range(sent_start, sent_end): + for sent_pos in range(sent_start, sent_end): sent_len += ONEF - sent_row = c.sent_adresses[i] * size + sent_row = c.sent_adresses[sent_pos] * size - word_idx = c.word_indices[i] - ngrams = c.subwords_idx_len[i] + word_idx = c.word_indices[sent_pos] + ngrams = c.subwords_idx_len[sent_pos] if ngrams == 0: - word_row = c.word_indices[i] * size + word_row = c.word_indices[sent_pos] * size swrmax_pool( &size, @@ -252,8 +252,8 @@ cdef void compute_ft_sentence_pooling( else: memset(c.mem, 0, size * cython.sizeof(REAL_t)) inv_ngram = (ONEF / ngrams) * c.oov_weight - for j in range(ngrams): - ngram_row = c.subwords_idx[(i * MAX_NGRAMS)+j] * size + for ngram_pos in range(ngrams): + ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size saxpy( &size, &inv_ngram, @@ -271,6 +271,121 @@ cdef void compute_ft_sentence_pooling( ) # There's nothing to do here for many-to-one mappings +cdef void compute_ft_sentence_hier_pooling( + FTSentenceVecsConfig *c, + uINT_t num_sentences, + uINT_t window_size, +) nogil: + """Perform optimized sentence-level hierarchical max pooling for FastText model. + + Parameters + ---------- + c : FTSentenceVecsConfig * + A pointer to a fully initialized and populated struct. + num_sentences : uINT_t + The number of sentences used to train the model. + window_size : uINT_t + The local window size. + + Notes + ----- + This routine DOES provide oov support. + + """ + # The naming of the i,j,k vars is a bit different here + + cdef: + int size = c.size + + uINT_t sent_idx, sent_start, sent_end, sent_row + + uINT_t ngram_row, ngrams + + uINT_t sent_pos, window_pos, ngram_pos, word_idx, word_row + + REAL_t sent_len, win_len + REAL_t inv_count, inv_ngram + REAL_t oov_weight = c.oov_weight + + for sent_idx in range(num_sentences): + sent_start = c.sentence_boundary[sent_idx] + sent_end = c.sentence_boundary[sent_idx + 1] + sent_len = ZEROF + + for sent_pos in range(sent_start, sent_end): + sent_len += ONEF + sent_row = c.sent_adresses[sent_pos] * size + + if sent_pos + window_size > sent_end: + window_end = sent_end + else: + window_end = sent_pos + window_size + + # Compute the locally averaged window + win_len = ZEROF + memset(c.mem, 0, size * cython.sizeof(REAL_t)) + memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + for window_pos in range(sent_pos, window_end): + win_len += ONEF + ngrams = c.subwords_idx_len[window_pos] + + if ngrams == 0: + word_row = c.word_indices[window_pos] * size + word_idx = c.word_indices[window_pos] + + saxpy( + &size, + &c.word_weights[word_idx], + &c.word_vectors[word_row], + &ONE, + c.mem, + &ONE + ) + + else: + memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + inv_ngram = (ONEF / ngrams) * c.oov_weight + for ngram_pos in range(ngrams): + ngram_row = c.subwords_idx[(window_pos * MAX_NGRAMS)+ngram_pos] * size + saxpy( + &size, + &inv_ngram, + &c.ngram_vectors[ngram_row], + &ONE, + c.mem2, + &ONE + ) + + saxpy( + &size, + &oov_weight, + c.mem2, + &ONE, + c.mem, + &ONE + ) + + memset(c.mem2, 0, size * cython.sizeof(REAL_t)) + # Rescale for dynamic window size + if win_len > ZEROF: + inv_count = ONEF / win_len + saxpy( + &size, + &inv_count, + c.mem, + &ONE, + c.mem2, + &ONE + ) + + swrmax_pool( + &size, + &ONEF, + c.mem2, + &c.sentence_vectors[sent_row], + ) + # There's nothing to do here for many-to-one mappings + def train_pooling_cy( model, indexed_sentences, @@ -340,6 +455,13 @@ def train_pooling_cy( if not model.hierarchical: with nogil: compute_ft_sentence_pooling(&ft, eff_sentences) + else: + with nogil: + compute_ft_sentence_hier_pooling( + &ft, + eff_sentences, + window_size, + ) return eff_sentences, eff_words diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index 74399e2..744210b 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -105,7 +105,7 @@ def test_do_train_job(self): ) self.assertEqual((105, DIM), self.model.sv.vectors.shape) - def test_pooling_train_np_w2v(self): + def test_pool_train_np_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() @@ -180,8 +180,8 @@ def test_pool_cy_equal_np_w2v_random(self): o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) - self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - + self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + def test_pool_train_np_ft(self): m = MaxPooling(FT) m.prep.prepare_vectors( @@ -222,7 +222,7 @@ def test_pool_train_cy_ft(self): np.allclose(1080970.2, m.sv[3]) ) - def test_cy_equal_np_ft_random(self): + def test_pool_cy_equal_np_ft_random(self): ft = FastText(size=20, min_count=1) ft.build_vocab(SENTENCES) @@ -252,21 +252,18 @@ def test_cy_equal_np_ft_random(self): self.assertEqual(o1, o2) self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - def test_pooling_train_np_w2v_non_negative(self): + def test_pool_np_w2v_non_negative(self): mpool = MaxPooling(W2V_R) mpool.train(self.sentences) self.assertTrue((mpool.sv.vectors >= 0).all()) - def test_hpooling_train_np_w2v_non_negative(self): - mpool = MaxPooling(W2V_R, hierarchical=True) - mpool.train(self.sentences) - self.assertTrue((mpool.sv.vectors >= 0).all()) - - def test_pooling_train_np_ft_non_negative(self): + def test_pool_np_ft_non_negative(self): mpool = MaxPooling(FT_R) mpool.train(self.sentences) self.assertTrue((mpool.sv.vectors >= 0).all()) + ### Hierarchical Tests start here + def test_hier_pooling_train_np_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() @@ -299,23 +296,20 @@ def test_hier_pooling_train_cy_w2v(self): self.assertTrue((183 == self.model.sv[0]).all()) self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) - def test_hier_pooling_train_np_ft(self): - m = MaxPooling(FT) + m = MaxPooling(FT, hierarchical=True) m.prep.prepare_vectors( sv=m.sv, total_sentences=len(self.sentences), update=False ) m._pre_train_calls() mem = m._get_thread_working_mem() - m.hierarchical = True - output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) self.assertEqual((5, 19), output) self.assertTrue((183 == m.sv[0]).all()) - self.assertTrue((737413.9 == m.sv[2]).all()) - self.assertTrue((1080970.2 == m.sv[3]).all()) + self.assertTrue(np.allclose(737413.9, m.sv[2])) + self.assertTrue(np.allclose(1080970.2, m.sv[3])) """ Note to future self: Due to the size of the ngram vectors, @@ -324,7 +318,55 @@ def test_hier_pooling_train_np_ft(self): TODO: This unittest is thus a bit flawed. Maybe fix? """ - def test_hier_pooling_train_np_ft_non_negative(self): + def test_hier_pooling_train_cy_ft(self): + m = MaxPooling(FT, hierarchical=True) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + mem = m._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + self.assertTrue((183 == m.sv[0]).all()) + self.assertTrue(np.allclose(737413.9, m.sv[2])) + self.assertTrue(np.allclose(1080970.2, m.sv[3])) + + def test_hier_pool_cy_equal_np_w2v_random(self): + w2v = Word2Vec(min_count=1, size=DIM) + # Random initialization + w2v.build_vocab(SENTENCES) + + m1 = MaxPooling(w2v, hierarchical=True) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + + m2 = MaxPooling(w2v, hierarchical=True) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + + self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + def test_hier_pool_cy_w2v_non_negative(self): + mpool = MaxPooling(W2V_R, hierarchical=True) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) + + def test_hier_pool_cy_ft_non_negative(self): mpool = MaxPooling(FT_R, hierarchical=True) mpool.train(self.sentences) self.assertTrue((mpool.sv.vectors >= 0).all()) From 65c94f8c5220c310e5a7ee10a755b92cb13f21fe Mon Sep 17 00:00:00 2001 From: OliverB Date: Mon, 2 Mar 2020 18:24:46 +0100 Subject: [PATCH 34/63] Minor syntax changes --- fse/models/pooling.py | 9 +- fse/models/pooling_inner.c | 264 ++++++++++++++++++------------------- 2 files changed, 139 insertions(+), 134 deletions(-) diff --git a/fse/models/pooling.py b/fse/models/pooling.py index 297f7f3..09616c1 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -156,7 +156,9 @@ def get_ft_vector(word: str) -> ndarray: sent = obj[0] sent_adr = obj[1] - word_indices = [vocab[word].index for word in sent if word in vocab] + word_indices = [ + vocab[word].index for word in sent if word in vocab + ] eff_sentences += 1 if not len(word_indices): continue @@ -165,7 +167,10 @@ def get_ft_vector(word: str) -> ndarray: if not hierarchical: # Take the maxium value along the axis mem = np_amax( - np_mult(w_vectors[word_indices], w_weights[word_indices][:, None]), + np_mult( + w_vectors[word_indices], + w_weights[word_indices][:, None] + ), axis=0, ) s_vectors[sent_adr] = np_maximum(s_vectors[sent_adr], mem,) diff --git a/fse/models/pooling_inner.c b/fse/models/pooling_inner.c index c6373cf..c4154bd 100644 --- a/fse/models/pooling_inner.c +++ b/fse/models/pooling_inner.c @@ -2593,8 +2593,8 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 /* function exit code */ } -/* "pooling_inner.pyx":275 - * +/* "pooling_inner.pyx":274 + * # There's nothing to do here for many-to-one mappings * * cdef void compute_ft_sentence_hier_pooling( # <<<<<<<<<<<<<< * FTSentenceVecsConfig *c, @@ -2636,7 +2636,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_13average_inner_uINT_t __pyx_t_14; __pyx_t_13average_inner_uINT_t __pyx_t_15; - /* "pooling_inner.pyx":299 + /* "pooling_inner.pyx":298 * * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -2646,7 +2646,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":309 + /* "pooling_inner.pyx":308 * REAL_t sent_len, win_len * REAL_t inv_count, inv_ngram * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< @@ -2656,7 +2656,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_2 = __pyx_v_c->oov_weight; __pyx_v_oov_weight = __pyx_t_2; - /* "pooling_inner.pyx":311 + /* "pooling_inner.pyx":310 * REAL_t oov_weight = c.oov_weight * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -2668,7 +2668,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { __pyx_v_sent_idx = __pyx_t_5; - /* "pooling_inner.pyx":312 + /* "pooling_inner.pyx":311 * * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -2677,7 +2677,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":313 + /* "pooling_inner.pyx":312 * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -2686,7 +2686,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":314 + /* "pooling_inner.pyx":313 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -2695,7 +2695,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":316 + /* "pooling_inner.pyx":315 * sent_len = ZEROF * * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -2707,7 +2707,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { __pyx_v_sent_pos = __pyx_t_8; - /* "pooling_inner.pyx":317 + /* "pooling_inner.pyx":316 * * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -2716,7 +2716,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":318 + /* "pooling_inner.pyx":317 * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< @@ -2725,7 +2725,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":320 + /* "pooling_inner.pyx":319 * sent_row = c.sent_adresses[sent_pos] * size * * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< @@ -2735,7 +2735,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = (((__pyx_v_sent_pos + __pyx_v_window_size) > __pyx_v_sent_end) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":321 + /* "pooling_inner.pyx":320 * * if sent_pos + window_size > sent_end: * window_end = sent_end # <<<<<<<<<<<<<< @@ -2744,7 +2744,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_window_end = __pyx_v_sent_end; - /* "pooling_inner.pyx":320 + /* "pooling_inner.pyx":319 * sent_row = c.sent_adresses[sent_pos] * size * * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< @@ -2754,7 +2754,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py goto __pyx_L7; } - /* "pooling_inner.pyx":323 + /* "pooling_inner.pyx":322 * window_end = sent_end * else: * window_end = sent_pos + window_size # <<<<<<<<<<<<<< @@ -2766,7 +2766,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py } __pyx_L7:; - /* "pooling_inner.pyx":326 + /* "pooling_inner.pyx":325 * * # Compute the locally averaged window * win_len = ZEROF # <<<<<<<<<<<<<< @@ -2775,7 +2775,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_win_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":327 + /* "pooling_inner.pyx":326 * # Compute the locally averaged window * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2784,7 +2784,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":328 + /* "pooling_inner.pyx":327 * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2793,7 +2793,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":329 + /* "pooling_inner.pyx":328 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * for window_pos in range(sent_pos, window_end): # <<<<<<<<<<<<<< @@ -2805,7 +2805,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_12 = __pyx_v_sent_pos; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { __pyx_v_window_pos = __pyx_t_12; - /* "pooling_inner.pyx":330 + /* "pooling_inner.pyx":329 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * for window_pos in range(sent_pos, window_end): * win_len += ONEF # <<<<<<<<<<<<<< @@ -2814,7 +2814,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_win_len = (__pyx_v_win_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":331 + /* "pooling_inner.pyx":330 * for window_pos in range(sent_pos, window_end): * win_len += ONEF * ngrams = c.subwords_idx_len[window_pos] # <<<<<<<<<<<<<< @@ -2823,7 +2823,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_window_pos]); - /* "pooling_inner.pyx":333 + /* "pooling_inner.pyx":332 * ngrams = c.subwords_idx_len[window_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2833,7 +2833,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":334 + /* "pooling_inner.pyx":333 * * if ngrams == 0: * word_row = c.word_indices[window_pos] * size # <<<<<<<<<<<<<< @@ -2842,7 +2842,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_window_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":335 + /* "pooling_inner.pyx":334 * if ngrams == 0: * word_row = c.word_indices[window_pos] * size * word_idx = c.word_indices[window_pos] # <<<<<<<<<<<<<< @@ -2851,7 +2851,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_window_pos]); - /* "pooling_inner.pyx":337 + /* "pooling_inner.pyx":336 * word_idx = c.word_indices[window_pos] * * saxpy( # <<<<<<<<<<<<<< @@ -2860,7 +2860,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); - /* "pooling_inner.pyx":333 + /* "pooling_inner.pyx":332 * ngrams = c.subwords_idx_len[window_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2870,7 +2870,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py goto __pyx_L10; } - /* "pooling_inner.pyx":347 + /* "pooling_inner.pyx":346 * * else: * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2880,7 +2880,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py /*else*/ { (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":348 + /* "pooling_inner.pyx":347 * else: * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< @@ -2889,7 +2889,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); - /* "pooling_inner.pyx":349 + /* "pooling_inner.pyx":348 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): # <<<<<<<<<<<<<< @@ -2901,7 +2901,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_15 = 0; __pyx_t_15 < __pyx_t_14; __pyx_t_15+=1) { __pyx_v_ngram_pos = __pyx_t_15; - /* "pooling_inner.pyx":350 + /* "pooling_inner.pyx":349 * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(window_pos * MAX_NGRAMS)+ngram_pos] * size # <<<<<<<<<<<<<< @@ -2910,7 +2910,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_window_pos * 40) + __pyx_v_ngram_pos)]) * __pyx_v_size); - /* "pooling_inner.pyx":351 + /* "pooling_inner.pyx":350 * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(window_pos * MAX_NGRAMS)+ngram_pos] * size * saxpy( # <<<<<<<<<<<<<< @@ -2920,7 +2920,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_ngram), (&(__pyx_v_c->ngram_vectors[__pyx_v_ngram_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE)); } - /* "pooling_inner.pyx":360 + /* "pooling_inner.pyx":359 * ) * * saxpy( # <<<<<<<<<<<<<< @@ -2932,7 +2932,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_L10:; } - /* "pooling_inner.pyx":369 + /* "pooling_inner.pyx":368 * ) * * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2941,7 +2941,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":371 + /* "pooling_inner.pyx":370 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * # Rescale for dynamic window size * if win_len > ZEROF: # <<<<<<<<<<<<<< @@ -2951,7 +2951,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = ((__pyx_v_win_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":372 + /* "pooling_inner.pyx":371 * # Rescale for dynamic window size * if win_len > ZEROF: * inv_count = ONEF / win_len # <<<<<<<<<<<<<< @@ -2960,7 +2960,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_win_len); - /* "pooling_inner.pyx":373 + /* "pooling_inner.pyx":372 * if win_len > ZEROF: * inv_count = ONEF / win_len * saxpy( # <<<<<<<<<<<<<< @@ -2969,7 +2969,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE)); - /* "pooling_inner.pyx":371 + /* "pooling_inner.pyx":370 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * # Rescale for dynamic window size * if win_len > ZEROF: # <<<<<<<<<<<<<< @@ -2978,7 +2978,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ } - /* "pooling_inner.pyx":382 + /* "pooling_inner.pyx":381 * ) * * swrmax_pool( # <<<<<<<<<<<<<< @@ -2989,8 +2989,8 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py } } - /* "pooling_inner.pyx":275 - * + /* "pooling_inner.pyx":274 + * # There's nothing to do here for many-to-one mappings * * cdef void compute_ft_sentence_hier_pooling( # <<<<<<<<<<<<<< * FTSentenceVecsConfig *c, @@ -3000,7 +3000,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py /* function exit code */ } -/* "pooling_inner.pyx":390 +/* "pooling_inner.pyx":389 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< @@ -3047,23 +3047,23 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self case 1: if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 390, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 389, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 2: if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 390, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 389, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 3: if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 390, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 389, __pyx_L3_error) } } if (unlikely(kw_args > 0)) { - if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 390, __pyx_L3_error) + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 389, __pyx_L3_error) } } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { goto __pyx_L5_argtuple_error; @@ -3080,7 +3080,7 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self } goto __pyx_L4_argument_unpacking_done; __pyx_L5_argtuple_error:; - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 390, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 389, __pyx_L3_error) __pyx_L3_error:; __Pyx_AddTraceback("pooling_inner.train_pooling_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __Pyx_RefNannyFinishContext(); @@ -3112,7 +3112,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_13average_inner_uINT_t __pyx_t_9; __Pyx_RefNannySetupContext("train_pooling_cy", 0); - /* "pooling_inner.pyx":419 + /* "pooling_inner.pyx":418 * """ * * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< @@ -3121,7 +3121,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_sentences = 0; - /* "pooling_inner.pyx":420 + /* "pooling_inner.pyx":419 * * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< @@ -3130,65 +3130,65 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_words = 0; - /* "pooling_inner.pyx":421 + /* "pooling_inner.pyx":420 * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 * cdef uINT_t window_size = model.window_size # <<<<<<<<<<<<<< * cdef BaseSentenceVecsConfig w2v * cdef FTSentenceVecsConfig ft */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_size); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 421, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_size); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 420, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_1); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 421, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_1); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 420, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_v_window_size = ((__pyx_t_13average_inner_uINT_t)__pyx_t_2); - /* "pooling_inner.pyx":425 + /* "pooling_inner.pyx":424 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< * init_base_s2v_config(&w2v, model, target, memory) * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 425, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 424, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 425, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 424, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_4 = ((!__pyx_t_3) != 0); if (__pyx_t_4) { - /* "pooling_inner.pyx":426 + /* "pooling_inner.pyx":425 * * if not model.is_ft: * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< * * eff_sentences, eff_words = populate_base_s2v_config( */ - __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 426, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 425, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":430 + /* "pooling_inner.pyx":429 * eff_sentences, eff_words = populate_base_s2v_config( * &w2v, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 430, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 429, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 430, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 429, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":428 + /* "pooling_inner.pyx":427 * init_base_s2v_config(&w2v, model, target, memory) * * eff_sentences, eff_words = populate_base_s2v_config( # <<<<<<<<<<<<<< * &w2v, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 428, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 427, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -3197,7 +3197,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 428, __pyx_L1_error) + __PYX_ERR(0, 427, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -3210,15 +3210,15 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_5); __Pyx_INCREF(__pyx_t_6); #else - __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 428, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 427, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); - __pyx_t_6 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 428, __pyx_L1_error) + __pyx_t_6 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 427, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 428, __pyx_L1_error) + __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 427, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_8 = Py_TYPE(__pyx_t_7)->tp_iternext; @@ -3226,7 +3226,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_5); index = 1; __pyx_t_6 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_6)) goto __pyx_L4_unpacking_failed; __Pyx_GOTREF(__pyx_t_6); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 428, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 427, __pyx_L1_error) __pyx_t_8 = NULL; __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; goto __pyx_L5_unpacking_done; @@ -3234,31 +3234,31 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; __pyx_t_8 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 428, __pyx_L1_error) + __PYX_ERR(0, 427, __pyx_L1_error) __pyx_L5_unpacking_done:; } - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 428, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 427, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 428, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 427, __pyx_L1_error) __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_v_eff_sentences = __pyx_t_2; __pyx_v_eff_words = __pyx_t_9; - /* "pooling_inner.pyx":434 + /* "pooling_inner.pyx":433 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: * compute_base_sentence_pooling( */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 434, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 433, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 434, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 433, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_3 = ((!__pyx_t_4) != 0); if (__pyx_t_3) { - /* "pooling_inner.pyx":435 + /* "pooling_inner.pyx":434 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -3273,7 +3273,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":436 + /* "pooling_inner.pyx":435 * if not model.hierarchical: * with nogil: * compute_base_sentence_pooling( # <<<<<<<<<<<<<< @@ -3283,7 +3283,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_base_sentence_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":435 + /* "pooling_inner.pyx":434 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -3302,7 +3302,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } } - /* "pooling_inner.pyx":434 + /* "pooling_inner.pyx":433 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< @@ -3312,7 +3312,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L6; } - /* "pooling_inner.pyx":441 + /* "pooling_inner.pyx":440 * ) * else: * with nogil: # <<<<<<<<<<<<<< @@ -3328,7 +3328,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":442 + /* "pooling_inner.pyx":441 * else: * with nogil: * compute_base_sentence_hier_pooling( # <<<<<<<<<<<<<< @@ -3338,7 +3338,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences, __pyx_v_window_size); } - /* "pooling_inner.pyx":441 + /* "pooling_inner.pyx":440 * ) * else: * with nogil: # <<<<<<<<<<<<<< @@ -3359,7 +3359,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } __pyx_L6:; - /* "pooling_inner.pyx":425 + /* "pooling_inner.pyx":424 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< @@ -3369,7 +3369,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L3; } - /* "pooling_inner.pyx":448 + /* "pooling_inner.pyx":447 * ) * else: * init_ft_s2v_config(&ft, model, target, memory) # <<<<<<<<<<<<<< @@ -3377,31 +3377,31 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec * eff_sentences, eff_words = populate_ft_s2v_config( */ /*else*/ { - __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 448, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 447, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":452 + /* "pooling_inner.pyx":451 * eff_sentences, eff_words = populate_ft_s2v_config( * &ft, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 452, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 451, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_6 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 452, __pyx_L1_error) + __pyx_t_6 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 451, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":450 + /* "pooling_inner.pyx":449 * init_ft_s2v_config(&ft, model, target, memory) * * eff_sentences, eff_words = populate_ft_s2v_config( # <<<<<<<<<<<<<< * &ft, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_6, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 450, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_6, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 449, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -3410,7 +3410,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 450, __pyx_L1_error) + __PYX_ERR(0, 449, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -3423,15 +3423,15 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_6); __Pyx_INCREF(__pyx_t_5); #else - __pyx_t_6 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 450, __pyx_L1_error) + __pyx_t_6 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 449, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); - __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 450, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 449, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 450, __pyx_L1_error) + __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 449, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_8 = Py_TYPE(__pyx_t_7)->tp_iternext; @@ -3439,7 +3439,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_6); index = 1; __pyx_t_5 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_5)) goto __pyx_L13_unpacking_failed; __Pyx_GOTREF(__pyx_t_5); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 450, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 449, __pyx_L1_error) __pyx_t_8 = NULL; __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; goto __pyx_L14_unpacking_done; @@ -3447,31 +3447,31 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; __pyx_t_8 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 450, __pyx_L1_error) + __PYX_ERR(0, 449, __pyx_L1_error) __pyx_L14_unpacking_done:; } - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 450, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 449, __pyx_L1_error) __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 450, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 449, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; __pyx_v_eff_sentences = __pyx_t_9; __pyx_v_eff_words = __pyx_t_2; - /* "pooling_inner.pyx":456 + /* "pooling_inner.pyx":455 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: * compute_ft_sentence_pooling(&ft, eff_sentences) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 456, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 455, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 456, __pyx_L1_error) + __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 455, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_4 = ((!__pyx_t_3) != 0); if (__pyx_t_4) { - /* "pooling_inner.pyx":457 + /* "pooling_inner.pyx":456 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -3486,7 +3486,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":458 + /* "pooling_inner.pyx":457 * if not model.hierarchical: * with nogil: * compute_ft_sentence_pooling(&ft, eff_sentences) # <<<<<<<<<<<<<< @@ -3496,7 +3496,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_ft_sentence_pooling((&__pyx_v_ft), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":457 + /* "pooling_inner.pyx":456 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -3515,7 +3515,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } } - /* "pooling_inner.pyx":456 + /* "pooling_inner.pyx":455 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< @@ -3525,7 +3525,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L15; } - /* "pooling_inner.pyx":460 + /* "pooling_inner.pyx":459 * compute_ft_sentence_pooling(&ft, eff_sentences) * else: * with nogil: # <<<<<<<<<<<<<< @@ -3541,7 +3541,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":461 + /* "pooling_inner.pyx":460 * else: * with nogil: * compute_ft_sentence_hier_pooling( # <<<<<<<<<<<<<< @@ -3551,7 +3551,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling((&__pyx_v_ft), __pyx_v_eff_sentences, __pyx_v_window_size); } - /* "pooling_inner.pyx":460 + /* "pooling_inner.pyx":459 * compute_ft_sentence_pooling(&ft, eff_sentences) * else: * with nogil: # <<<<<<<<<<<<<< @@ -3574,7 +3574,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } __pyx_L3:; - /* "pooling_inner.pyx":467 + /* "pooling_inner.pyx":466 * ) * * return eff_sentences, eff_words # <<<<<<<<<<<<<< @@ -3582,11 +3582,11 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec * def init(): */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 467, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 466, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 467, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 466, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); - __pyx_t_6 = PyTuple_New(2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 467, __pyx_L1_error) + __pyx_t_6 = PyTuple_New(2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 466, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_6, 0, __pyx_t_1); @@ -3598,7 +3598,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_6 = 0; goto __pyx_L0; - /* "pooling_inner.pyx":390 + /* "pooling_inner.pyx":389 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< @@ -3620,7 +3620,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec return __pyx_r; } -/* "pooling_inner.pyx":469 +/* "pooling_inner.pyx":468 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -3648,7 +3648,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("init", 0); - /* "pooling_inner.pyx":470 + /* "pooling_inner.pyx":469 * * def init(): * return 1 # <<<<<<<<<<<<<< @@ -3660,7 +3660,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __pyx_r = __pyx_int_1; goto __pyx_L0; - /* "pooling_inner.pyx":469 + /* "pooling_inner.pyx":468 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -6274,26 +6274,26 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { __Pyx_GOTREF(__pyx_tuple__7); __Pyx_GIVEREF(__pyx_tuple__7); - /* "pooling_inner.pyx":390 + /* "pooling_inner.pyx":389 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_tuple__9 = PyTuple_Pack(9, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_window_size, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 390, __pyx_L1_error) + __pyx_tuple__9 = PyTuple_Pack(9, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_window_size, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 389, __pyx_L1_error) __Pyx_GOTREF(__pyx_tuple__9); __Pyx_GIVEREF(__pyx_tuple__9); - __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 9, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 390, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 390, __pyx_L1_error) + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 9, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 389, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 389, __pyx_L1_error) - /* "pooling_inner.pyx":469 + /* "pooling_inner.pyx":468 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 469, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 469, __pyx_L1_error) + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 468, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 468, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -6656,58 +6656,58 @@ if (!__Pyx_RefNanny) { if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_2) < 0) __PYX_ERR(0, 22, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":390 + /* "pooling_inner.pyx":389 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 390, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 389, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 390, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 389, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":469 + /* "pooling_inner.pyx":468 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 469, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 468, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 469, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 468, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":472 + /* "pooling_inner.pyx":471 * return 1 * * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 472, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 471, __pyx_L1_error) - /* "pooling_inner.pyx":473 + /* "pooling_inner.pyx":472 * * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 473, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 472, __pyx_L1_error) - /* "pooling_inner.pyx":474 + /* "pooling_inner.pyx":473 * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() # <<<<<<<<<<<<<< */ - __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 474, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 473, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 474, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 473, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 474, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 473, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; /* "pooling_inner.pyx":1 From bebde9d050982c8aa9fd914466dde4f763da25fe Mon Sep 17 00:00:00 2001 From: OliverB Date: Mon, 2 Mar 2020 19:53:39 +0100 Subject: [PATCH 35/63] Working hpool + stride all models --- fse/models/pooling.py | 22 +- fse/models/pooling_inner.c | 584 ++++++++++++++++++++--------------- fse/models/pooling_inner.pyx | 22 +- fse/test/test_pooling.py | 263 +++++++++++++--- 4 files changed, 585 insertions(+), 306 deletions(-) diff --git a/fse/models/pooling.py b/fse/models/pooling.py index 09616c1..2a1ebde 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -102,7 +102,8 @@ def train_pooling_np( mem = memory[0] hierarchical = model.hierarchical - window = model.window_size + window_size = model.window_size + window_stride = model.window_stride if is_ft: # NOTE: For Fasttext: Use wv.vectors_vocab @@ -177,8 +178,10 @@ def get_ft_vector(word: str) -> ndarray: else: # More expensive iteration for word_index, _ in enumerate(word_indices): + if word_index % window_stride != 0: + continue # Compute the local window - window_indices = word_indices[word_index : word_index + window] + window_indices = word_indices[word_index : word_index + window_size] # Perform average pooling [0,1,2,3,4] mem = np_sum( np_mult( @@ -209,18 +212,22 @@ def get_ft_vector(word: str) -> ndarray: get_ft_vector(word), s_vectors[sent_adr], ) else: + if sent_adr == 4: + print("hi") + for word_index, word in enumerate(sent): + if word_index % window_stride != 0: + continue mem.fill(0.0) mem += get_ft_vector(word) count = 1 - for context in sent[word_index : word_index + window]: + for context in sent[word_index : word_index + window_size]: if word == context: continue mem += get_ft_vector(context) count += 1 mem /= count - s_vectors[sent_adr] = np_maximum(mem, s_vectors[sent_adr],) return eff_sentences, eff_words @@ -270,10 +277,10 @@ def __init__( model: BaseKeyedVectors, hierarchical: bool = False, window_size: int = 5, + window_stride: int = 1, sv_mapfile_path: str = None, wv_mapfile_path: str = None, workers: int = 1, - **kwargs ): """ Max pooling sentence embeddings model. Performs a simple maximum pooling operation over all words in a sentences without further transformations. @@ -290,6 +297,8 @@ def __init__( If true, then perform hierarchical pooling operation window_size : int Set the size of the window used for hierarchical pooling operation + window_stride: int + Set adjacency of the window used for hierarchical pooling operation sv_mapfile_path : str, optional Optional path to store the sentence-vectors in for very large datasets. Used for memmap. wv_mapfile_path : str, optional @@ -301,6 +310,7 @@ def __init__( """ self.hierarchical = bool(hierarchical) self.window_size = int(window_size) + self.window_stride = int(window_stride) super(MaxPooling, self).__init__( model=model, @@ -328,6 +338,8 @@ def _check_parameter_sanity(self, **kwargs): raise ValueError("All word weights must equal one for pool") if self.window_size < 1: raise ValueError("Window size must be greater than 1") + if not 1 <= self.window_stride <= self.window_size: + raise ValueError(f"Window stride must be 1 <= stride <= {self.window_size}") def _pre_train_calls(self, **kwargs): """Function calls to perform before training """ diff --git a/fse/models/pooling_inner.c b/fse/models/pooling_inner.c index c4154bd..f8d4a14 100644 --- a/fse/models/pooling_inner.c +++ b/fse/models/pooling_inner.c @@ -1725,9 +1725,9 @@ static PyObject *(*__pyx_f_13average_inner_populate_ft_s2v_config)(struct __pyx_ /* Module declarations from 'pooling_inner' */ static void __pyx_f_13pooling_inner_swrmax_pool(int const *, float const *, float const *, float *); /*proto*/ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ -static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_uINT_t); /*proto*/ +static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_REAL_t); /*proto*/ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ -static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_uINT_t); /*proto*/ +static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_REAL_t); /*proto*/ #define __Pyx_MODULE_NAME "pooling_inner" extern int __pyx_module_is_main_pooling_inner; int __pyx_module_is_main_pooling_inner = 0; @@ -1764,6 +1764,7 @@ static const char __pyx_k_RuntimeError[] = "RuntimeError"; static const char __pyx_k_hierarchical[] = "hierarchical"; static const char __pyx_k_eff_sentences[] = "eff_sentences"; static const char __pyx_k_pooling_inner[] = "pooling_inner"; +static const char __pyx_k_window_stride[] = "window_stride"; static const char __pyx_k_train_pooling_cy[] = "train_pooling_cy"; static const char __pyx_k_indexed_sentences[] = "indexed_sentences"; static const char __pyx_k_pooling_inner_pyx[] = "pooling_inner.pyx"; @@ -1821,6 +1822,7 @@ static PyObject *__pyx_kp_u_unknown_dtype_code_in_numpy_pxd; static PyObject *__pyx_n_s_vocab; static PyObject *__pyx_n_s_w2v; static PyObject *__pyx_n_s_window_size; +static PyObject *__pyx_n_s_window_stride; static PyObject *__pyx_n_s_wv; static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObject *__pyx_self, PyObject *__pyx_v_model, PyObject *__pyx_v_indexed_sentences, PyObject *__pyx_v_target, PyObject *__pyx_v_memory); /* proto */ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_self); /* proto */ @@ -2057,7 +2059,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t * uINT_t num_sentences, */ -static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences, __pyx_t_13average_inner_uINT_t __pyx_v_window_size) { +static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences, __pyx_t_13average_inner_uINT_t __pyx_v_window_size, __pyx_t_13average_inner_REAL_t __pyx_v_window_stride) { int __pyx_v_size; __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; @@ -2083,8 +2085,8 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ __pyx_t_13average_inner_uINT_t __pyx_t_10; __pyx_t_13average_inner_uINT_t __pyx_t_11; - /* "pooling_inner.pyx":139 - * + /* "pooling_inner.pyx":141 + * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< * @@ -2093,7 +2095,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":147 + /* "pooling_inner.pyx":149 * REAL_t sent_len, win_len, inv_count * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -2105,7 +2107,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) { __pyx_v_sent_idx = __pyx_t_4; - /* "pooling_inner.pyx":148 + /* "pooling_inner.pyx":150 * * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -2114,7 +2116,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":149 + /* "pooling_inner.pyx":151 * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -2123,7 +2125,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":150 + /* "pooling_inner.pyx":152 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -2132,37 +2134,65 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":152 + /* "pooling_inner.pyx":154 * sent_len = ZEROF * * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< * sent_len += ONEF - * sent_row = c.sent_adresses[sent_pos] * size + * */ __pyx_t_5 = __pyx_v_sent_end; __pyx_t_6 = __pyx_t_5; for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { __pyx_v_sent_pos = __pyx_t_7; - /* "pooling_inner.pyx":153 + /* "pooling_inner.pyx":155 * * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< - * sent_row = c.sent_adresses[sent_pos] * size * + * if (sent_len-ONEF) % window_stride != ZEROF: */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":154 - * for sent_pos in range(sent_start, sent_end): + /* "pooling_inner.pyx":157 + * sent_len += ONEF + * + * if (sent_len-ONEF) % window_stride != ZEROF: # <<<<<<<<<<<<<< + * continue + * + */ + __pyx_t_8 = ((fmodf((__pyx_v_sent_len - __pyx_v_13average_inner_ONEF), __pyx_v_window_stride) != __pyx_v_13average_inner_ZEROF) != 0); + if (__pyx_t_8) { + + /* "pooling_inner.pyx":158 + * + * if (sent_len-ONEF) % window_stride != ZEROF: + * continue # <<<<<<<<<<<<<< + * + * sent_row = c.sent_adresses[sent_pos] * size + */ + goto __pyx_L5_continue; + + /* "pooling_inner.pyx":157 * sent_len += ONEF + * + * if (sent_len-ONEF) % window_stride != ZEROF: # <<<<<<<<<<<<<< + * continue + * + */ + } + + /* "pooling_inner.pyx":160 + * continue + * * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< * * if sent_pos + window_size > sent_end: */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":156 + /* "pooling_inner.pyx":162 * sent_row = c.sent_adresses[sent_pos] * size * * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< @@ -2172,7 +2202,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ __pyx_t_8 = (((__pyx_v_sent_pos + __pyx_v_window_size) > __pyx_v_sent_end) != 0); if (__pyx_t_8) { - /* "pooling_inner.pyx":157 + /* "pooling_inner.pyx":163 * * if sent_pos + window_size > sent_end: * window_end = sent_end # <<<<<<<<<<<<<< @@ -2181,17 +2211,17 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_window_end = __pyx_v_sent_end; - /* "pooling_inner.pyx":156 + /* "pooling_inner.pyx":162 * sent_row = c.sent_adresses[sent_pos] * size * * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< * window_end = sent_end * else: */ - goto __pyx_L7; + goto __pyx_L8; } - /* "pooling_inner.pyx":159 + /* "pooling_inner.pyx":165 * window_end = sent_end * else: * window_end = sent_pos + window_size # <<<<<<<<<<<<<< @@ -2201,9 +2231,9 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ /*else*/ { __pyx_v_window_end = (__pyx_v_sent_pos + __pyx_v_window_size); } - __pyx_L7:; + __pyx_L8:; - /* "pooling_inner.pyx":162 + /* "pooling_inner.pyx":168 * * # Compute the locally averaged window * win_len = ZEROF # <<<<<<<<<<<<<< @@ -2212,7 +2242,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_win_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":163 + /* "pooling_inner.pyx":169 * # Compute the locally averaged window * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2221,7 +2251,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":164 + /* "pooling_inner.pyx":170 * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2230,7 +2260,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":165 + /* "pooling_inner.pyx":171 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * for window_pos in range(sent_pos, window_end): # <<<<<<<<<<<<<< @@ -2242,7 +2272,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ for (__pyx_t_11 = __pyx_v_sent_pos; __pyx_t_11 < __pyx_t_10; __pyx_t_11+=1) { __pyx_v_window_pos = __pyx_t_11; - /* "pooling_inner.pyx":166 + /* "pooling_inner.pyx":172 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * for window_pos in range(sent_pos, window_end): * win_len += ONEF # <<<<<<<<<<<<<< @@ -2251,7 +2281,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_win_len = (__pyx_v_win_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":168 + /* "pooling_inner.pyx":174 * win_len += ONEF * * word_row = c.word_indices[window_pos] * size # <<<<<<<<<<<<<< @@ -2260,7 +2290,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_window_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":169 + /* "pooling_inner.pyx":175 * * word_row = c.word_indices[window_pos] * size * word_idx = c.word_indices[window_pos] # <<<<<<<<<<<<<< @@ -2269,7 +2299,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_window_pos]); - /* "pooling_inner.pyx":171 + /* "pooling_inner.pyx":177 * word_idx = c.word_indices[window_pos] * * saxpy( # <<<<<<<<<<<<<< @@ -2279,7 +2309,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); } - /* "pooling_inner.pyx":181 + /* "pooling_inner.pyx":187 * * # Rescale for dynamic window size * if win_len > ZEROF: # <<<<<<<<<<<<<< @@ -2289,7 +2319,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ __pyx_t_8 = ((__pyx_v_win_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_8) { - /* "pooling_inner.pyx":182 + /* "pooling_inner.pyx":188 * # Rescale for dynamic window size * if win_len > ZEROF: * inv_count = ONEF / win_len # <<<<<<<<<<<<<< @@ -2298,7 +2328,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_win_len); - /* "pooling_inner.pyx":183 + /* "pooling_inner.pyx":189 * if win_len > ZEROF: * inv_count = ONEF / win_len * saxpy( # <<<<<<<<<<<<<< @@ -2307,7 +2337,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE)); - /* "pooling_inner.pyx":181 + /* "pooling_inner.pyx":187 * * # Rescale for dynamic window size * if win_len > ZEROF: # <<<<<<<<<<<<<< @@ -2316,7 +2346,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ } - /* "pooling_inner.pyx":192 + /* "pooling_inner.pyx":198 * ) * * swrmax_pool( # <<<<<<<<<<<<<< @@ -2324,6 +2354,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ * &ONEF, */ __pyx_f_13pooling_inner_swrmax_pool((&__pyx_v_size), (&__pyx_v_13average_inner_ONEF), __pyx_v_c->mem2, (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row]))); + __pyx_L5_continue:; } } @@ -2338,7 +2369,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ /* function exit code */ } -/* "pooling_inner.pyx":199 +/* "pooling_inner.pyx":205 * ) * * cdef void compute_ft_sentence_pooling( # <<<<<<<<<<<<<< @@ -2374,7 +2405,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_13average_inner_uINT_t __pyx_t_11; __pyx_t_13average_inner_uINT_t __pyx_t_12; - /* "pooling_inner.pyx":218 + /* "pooling_inner.pyx":224 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -2384,7 +2415,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":228 + /* "pooling_inner.pyx":234 * REAL_t sent_len * REAL_t inv_count, inv_ngram * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< @@ -2394,7 +2425,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_2 = __pyx_v_c->oov_weight; __pyx_v_oov_weight = __pyx_t_2; - /* "pooling_inner.pyx":230 + /* "pooling_inner.pyx":236 * REAL_t oov_weight = c.oov_weight * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -2406,7 +2437,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { __pyx_v_sent_idx = __pyx_t_5; - /* "pooling_inner.pyx":231 + /* "pooling_inner.pyx":237 * * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -2415,7 +2446,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":232 + /* "pooling_inner.pyx":238 * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -2424,7 +2455,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":233 + /* "pooling_inner.pyx":239 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -2433,7 +2464,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":235 + /* "pooling_inner.pyx":241 * sent_len = ZEROF * * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -2445,7 +2476,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { __pyx_v_sent_pos = __pyx_t_8; - /* "pooling_inner.pyx":236 + /* "pooling_inner.pyx":242 * * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -2454,7 +2485,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":237 + /* "pooling_inner.pyx":243 * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< @@ -2463,7 +2494,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":239 + /* "pooling_inner.pyx":245 * sent_row = c.sent_adresses[sent_pos] * size * * word_idx = c.word_indices[sent_pos] # <<<<<<<<<<<<<< @@ -2472,7 +2503,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_sent_pos]); - /* "pooling_inner.pyx":240 + /* "pooling_inner.pyx":246 * * word_idx = c.word_indices[sent_pos] * ngrams = c.subwords_idx_len[sent_pos] # <<<<<<<<<<<<<< @@ -2481,7 +2512,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_sent_pos]); - /* "pooling_inner.pyx":242 + /* "pooling_inner.pyx":248 * ngrams = c.subwords_idx_len[sent_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2491,7 +2522,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":243 + /* "pooling_inner.pyx":249 * * if ngrams == 0: * word_row = c.word_indices[sent_pos] * size # <<<<<<<<<<<<<< @@ -2500,7 +2531,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_sent_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":245 + /* "pooling_inner.pyx":251 * word_row = c.word_indices[sent_pos] * size * * swrmax_pool( # <<<<<<<<<<<<<< @@ -2509,7 +2540,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_f_13pooling_inner_swrmax_pool((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row]))); - /* "pooling_inner.pyx":242 + /* "pooling_inner.pyx":248 * ngrams = c.subwords_idx_len[sent_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2519,7 +2550,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 goto __pyx_L7; } - /* "pooling_inner.pyx":253 + /* "pooling_inner.pyx":259 * * else: * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2529,7 +2560,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 /*else*/ { (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":254 + /* "pooling_inner.pyx":260 * else: * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< @@ -2538,7 +2569,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); - /* "pooling_inner.pyx":255 + /* "pooling_inner.pyx":261 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): # <<<<<<<<<<<<<< @@ -2550,7 +2581,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 for (__pyx_t_12 = 0; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { __pyx_v_ngram_pos = __pyx_t_12; - /* "pooling_inner.pyx":256 + /* "pooling_inner.pyx":262 * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size # <<<<<<<<<<<<<< @@ -2559,7 +2590,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_sent_pos * 40) + __pyx_v_ngram_pos)]) * __pyx_v_size); - /* "pooling_inner.pyx":257 + /* "pooling_inner.pyx":263 * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size * saxpy( # <<<<<<<<<<<<<< @@ -2569,7 +2600,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_ngram), (&(__pyx_v_c->ngram_vectors[__pyx_v_ngram_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); } - /* "pooling_inner.pyx":266 + /* "pooling_inner.pyx":272 * ) * * swrmax_pool( # <<<<<<<<<<<<<< @@ -2582,7 +2613,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 } } - /* "pooling_inner.pyx":199 + /* "pooling_inner.pyx":205 * ) * * cdef void compute_ft_sentence_pooling( # <<<<<<<<<<<<<< @@ -2593,7 +2624,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 /* function exit code */ } -/* "pooling_inner.pyx":274 +/* "pooling_inner.pyx":280 * # There's nothing to do here for many-to-one mappings * * cdef void compute_ft_sentence_hier_pooling( # <<<<<<<<<<<<<< @@ -2601,7 +2632,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 * uINT_t num_sentences, */ -static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences, __pyx_t_13average_inner_uINT_t __pyx_v_window_size) { +static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences, __pyx_t_13average_inner_uINT_t __pyx_v_window_size, __pyx_t_13average_inner_REAL_t __pyx_v_window_stride) { int __pyx_v_size; __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; @@ -2636,7 +2667,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_13average_inner_uINT_t __pyx_t_14; __pyx_t_13average_inner_uINT_t __pyx_t_15; - /* "pooling_inner.pyx":298 + /* "pooling_inner.pyx":307 * * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -2646,7 +2677,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":308 + /* "pooling_inner.pyx":317 * REAL_t sent_len, win_len * REAL_t inv_count, inv_ngram * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< @@ -2656,7 +2687,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_2 = __pyx_v_c->oov_weight; __pyx_v_oov_weight = __pyx_t_2; - /* "pooling_inner.pyx":310 + /* "pooling_inner.pyx":319 * REAL_t oov_weight = c.oov_weight * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -2668,7 +2699,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { __pyx_v_sent_idx = __pyx_t_5; - /* "pooling_inner.pyx":311 + /* "pooling_inner.pyx":320 * * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -2677,7 +2708,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":312 + /* "pooling_inner.pyx":321 * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -2686,7 +2717,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":313 + /* "pooling_inner.pyx":322 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -2695,37 +2726,65 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":315 + /* "pooling_inner.pyx":324 * sent_len = ZEROF * * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< * sent_len += ONEF - * sent_row = c.sent_adresses[sent_pos] * size + * */ __pyx_t_6 = __pyx_v_sent_end; __pyx_t_7 = __pyx_t_6; for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { __pyx_v_sent_pos = __pyx_t_8; - /* "pooling_inner.pyx":316 + /* "pooling_inner.pyx":325 * * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< - * sent_row = c.sent_adresses[sent_pos] * size * + * if (sent_len-ONEF) % window_stride != ZEROF: */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":317 - * for sent_pos in range(sent_start, sent_end): + /* "pooling_inner.pyx":327 * sent_len += ONEF + * + * if (sent_len-ONEF) % window_stride != ZEROF: # <<<<<<<<<<<<<< + * continue + * + */ + __pyx_t_9 = ((fmodf((__pyx_v_sent_len - __pyx_v_13average_inner_ONEF), __pyx_v_window_stride) != __pyx_v_13average_inner_ZEROF) != 0); + if (__pyx_t_9) { + + /* "pooling_inner.pyx":328 + * + * if (sent_len-ONEF) % window_stride != ZEROF: + * continue # <<<<<<<<<<<<<< + * + * sent_row = c.sent_adresses[sent_pos] * size + */ + goto __pyx_L5_continue; + + /* "pooling_inner.pyx":327 + * sent_len += ONEF + * + * if (sent_len-ONEF) % window_stride != ZEROF: # <<<<<<<<<<<<<< + * continue + * + */ + } + + /* "pooling_inner.pyx":330 + * continue + * * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< * * if sent_pos + window_size > sent_end: */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":319 + /* "pooling_inner.pyx":332 * sent_row = c.sent_adresses[sent_pos] * size * * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< @@ -2735,7 +2794,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = (((__pyx_v_sent_pos + __pyx_v_window_size) > __pyx_v_sent_end) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":320 + /* "pooling_inner.pyx":333 * * if sent_pos + window_size > sent_end: * window_end = sent_end # <<<<<<<<<<<<<< @@ -2744,17 +2803,17 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_window_end = __pyx_v_sent_end; - /* "pooling_inner.pyx":319 + /* "pooling_inner.pyx":332 * sent_row = c.sent_adresses[sent_pos] * size * * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< * window_end = sent_end * else: */ - goto __pyx_L7; + goto __pyx_L8; } - /* "pooling_inner.pyx":322 + /* "pooling_inner.pyx":335 * window_end = sent_end * else: * window_end = sent_pos + window_size # <<<<<<<<<<<<<< @@ -2764,9 +2823,9 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py /*else*/ { __pyx_v_window_end = (__pyx_v_sent_pos + __pyx_v_window_size); } - __pyx_L7:; + __pyx_L8:; - /* "pooling_inner.pyx":325 + /* "pooling_inner.pyx":338 * * # Compute the locally averaged window * win_len = ZEROF # <<<<<<<<<<<<<< @@ -2775,7 +2834,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_win_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":326 + /* "pooling_inner.pyx":339 * # Compute the locally averaged window * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2784,7 +2843,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":327 + /* "pooling_inner.pyx":340 * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2793,7 +2852,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":328 + /* "pooling_inner.pyx":341 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * for window_pos in range(sent_pos, window_end): # <<<<<<<<<<<<<< @@ -2805,7 +2864,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_12 = __pyx_v_sent_pos; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { __pyx_v_window_pos = __pyx_t_12; - /* "pooling_inner.pyx":329 + /* "pooling_inner.pyx":342 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * for window_pos in range(sent_pos, window_end): * win_len += ONEF # <<<<<<<<<<<<<< @@ -2814,7 +2873,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_win_len = (__pyx_v_win_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":330 + /* "pooling_inner.pyx":343 * for window_pos in range(sent_pos, window_end): * win_len += ONEF * ngrams = c.subwords_idx_len[window_pos] # <<<<<<<<<<<<<< @@ -2823,7 +2882,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_window_pos]); - /* "pooling_inner.pyx":332 + /* "pooling_inner.pyx":345 * ngrams = c.subwords_idx_len[window_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2833,7 +2892,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":333 + /* "pooling_inner.pyx":346 * * if ngrams == 0: * word_row = c.word_indices[window_pos] * size # <<<<<<<<<<<<<< @@ -2842,7 +2901,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_window_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":334 + /* "pooling_inner.pyx":347 * if ngrams == 0: * word_row = c.word_indices[window_pos] * size * word_idx = c.word_indices[window_pos] # <<<<<<<<<<<<<< @@ -2851,7 +2910,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_window_pos]); - /* "pooling_inner.pyx":336 + /* "pooling_inner.pyx":349 * word_idx = c.word_indices[window_pos] * * saxpy( # <<<<<<<<<<<<<< @@ -2860,17 +2919,17 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); - /* "pooling_inner.pyx":332 + /* "pooling_inner.pyx":345 * ngrams = c.subwords_idx_len[window_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< * word_row = c.word_indices[window_pos] * size * word_idx = c.word_indices[window_pos] */ - goto __pyx_L10; + goto __pyx_L11; } - /* "pooling_inner.pyx":346 + /* "pooling_inner.pyx":359 * * else: * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2880,7 +2939,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py /*else*/ { (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":347 + /* "pooling_inner.pyx":360 * else: * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< @@ -2889,7 +2948,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); - /* "pooling_inner.pyx":348 + /* "pooling_inner.pyx":361 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): # <<<<<<<<<<<<<< @@ -2901,7 +2960,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_15 = 0; __pyx_t_15 < __pyx_t_14; __pyx_t_15+=1) { __pyx_v_ngram_pos = __pyx_t_15; - /* "pooling_inner.pyx":349 + /* "pooling_inner.pyx":362 * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(window_pos * MAX_NGRAMS)+ngram_pos] * size # <<<<<<<<<<<<<< @@ -2910,7 +2969,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_window_pos * 40) + __pyx_v_ngram_pos)]) * __pyx_v_size); - /* "pooling_inner.pyx":350 + /* "pooling_inner.pyx":363 * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(window_pos * MAX_NGRAMS)+ngram_pos] * size * saxpy( # <<<<<<<<<<<<<< @@ -2920,7 +2979,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_ngram), (&(__pyx_v_c->ngram_vectors[__pyx_v_ngram_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE)); } - /* "pooling_inner.pyx":359 + /* "pooling_inner.pyx":372 * ) * * saxpy( # <<<<<<<<<<<<<< @@ -2929,10 +2988,10 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_oov_weight), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); } - __pyx_L10:; + __pyx_L11:; } - /* "pooling_inner.pyx":368 + /* "pooling_inner.pyx":381 * ) * * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2941,7 +3000,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":370 + /* "pooling_inner.pyx":383 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * # Rescale for dynamic window size * if win_len > ZEROF: # <<<<<<<<<<<<<< @@ -2951,7 +3010,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = ((__pyx_v_win_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":371 + /* "pooling_inner.pyx":384 * # Rescale for dynamic window size * if win_len > ZEROF: * inv_count = ONEF / win_len # <<<<<<<<<<<<<< @@ -2960,7 +3019,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_win_len); - /* "pooling_inner.pyx":372 + /* "pooling_inner.pyx":385 * if win_len > ZEROF: * inv_count = ONEF / win_len * saxpy( # <<<<<<<<<<<<<< @@ -2969,7 +3028,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE)); - /* "pooling_inner.pyx":370 + /* "pooling_inner.pyx":383 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * # Rescale for dynamic window size * if win_len > ZEROF: # <<<<<<<<<<<<<< @@ -2978,7 +3037,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ } - /* "pooling_inner.pyx":381 + /* "pooling_inner.pyx":394 * ) * * swrmax_pool( # <<<<<<<<<<<<<< @@ -2986,10 +3045,11 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py * &ONEF, */ __pyx_f_13pooling_inner_swrmax_pool((&__pyx_v_size), (&__pyx_v_13average_inner_ONEF), __pyx_v_c->mem2, (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row]))); + __pyx_L5_continue:; } } - /* "pooling_inner.pyx":274 + /* "pooling_inner.pyx":280 * # There's nothing to do here for many-to-one mappings * * cdef void compute_ft_sentence_hier_pooling( # <<<<<<<<<<<<<< @@ -3000,7 +3060,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py /* function exit code */ } -/* "pooling_inner.pyx":389 +/* "pooling_inner.pyx":402 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< @@ -3047,23 +3107,23 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self case 1: if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 389, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 402, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 2: if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 389, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 402, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 3: if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 389, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 402, __pyx_L3_error) } } if (unlikely(kw_args > 0)) { - if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 389, __pyx_L3_error) + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 402, __pyx_L3_error) } } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { goto __pyx_L5_argtuple_error; @@ -3080,7 +3140,7 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self } goto __pyx_L4_argument_unpacking_done; __pyx_L5_argtuple_error:; - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 389, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 402, __pyx_L3_error) __pyx_L3_error:; __Pyx_AddTraceback("pooling_inner.train_pooling_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __Pyx_RefNannyFinishContext(); @@ -3097,22 +3157,24 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_13average_inner_uINT_t __pyx_v_eff_sentences; __pyx_t_13average_inner_uINT_t __pyx_v_eff_words; __pyx_t_13average_inner_uINT_t __pyx_v_window_size; + __pyx_t_13average_inner_REAL_t __pyx_v_window_stride; struct __pyx_t_13average_inner_BaseSentenceVecsConfig __pyx_v_w2v; struct __pyx_t_13average_inner_FTSentenceVecsConfig __pyx_v_ft; PyObject *__pyx_r = NULL; __Pyx_RefNannyDeclarations PyObject *__pyx_t_1 = NULL; __pyx_t_13average_inner_uINT_t __pyx_t_2; - int __pyx_t_3; + __pyx_t_13average_inner_REAL_t __pyx_t_3; int __pyx_t_4; - PyObject *__pyx_t_5 = NULL; + int __pyx_t_5; PyObject *__pyx_t_6 = NULL; PyObject *__pyx_t_7 = NULL; - PyObject *(*__pyx_t_8)(PyObject *); - __pyx_t_13average_inner_uINT_t __pyx_t_9; + PyObject *__pyx_t_8 = NULL; + PyObject *(*__pyx_t_9)(PyObject *); + __pyx_t_13average_inner_uINT_t __pyx_t_10; __Pyx_RefNannySetupContext("train_pooling_cy", 0); - /* "pooling_inner.pyx":418 + /* "pooling_inner.pyx":431 * """ * * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< @@ -3121,144 +3183,157 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_sentences = 0; - /* "pooling_inner.pyx":419 + /* "pooling_inner.pyx":432 * * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< * cdef uINT_t window_size = model.window_size - * cdef BaseSentenceVecsConfig w2v + * cdef REAL_t window_stride = model.window_stride */ __pyx_v_eff_words = 0; - /* "pooling_inner.pyx":420 + /* "pooling_inner.pyx":433 * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 * cdef uINT_t window_size = model.window_size # <<<<<<<<<<<<<< + * cdef REAL_t window_stride = model.window_stride * cdef BaseSentenceVecsConfig w2v - * cdef FTSentenceVecsConfig ft */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_size); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 420, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_size); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 433, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_1); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 420, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_1); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 433, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_v_window_size = ((__pyx_t_13average_inner_uINT_t)__pyx_t_2); - /* "pooling_inner.pyx":424 + /* "pooling_inner.pyx":434 + * cdef uINT_t eff_words = 0 + * cdef uINT_t window_size = model.window_size + * cdef REAL_t window_stride = model.window_stride # <<<<<<<<<<<<<< + * cdef BaseSentenceVecsConfig w2v + * cdef FTSentenceVecsConfig ft + */ + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_stride); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 434, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_1); + __pyx_t_3 = __pyx_PyFloat_AsFloat(__pyx_t_1); if (unlikely((__pyx_t_3 == ((npy_float32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 434, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; + __pyx_v_window_stride = ((__pyx_t_13average_inner_REAL_t)__pyx_t_3); + + /* "pooling_inner.pyx":438 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< * init_base_s2v_config(&w2v, model, target, memory) * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 424, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 438, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 424, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 438, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_4 = ((!__pyx_t_3) != 0); - if (__pyx_t_4) { + __pyx_t_5 = ((!__pyx_t_4) != 0); + if (__pyx_t_5) { - /* "pooling_inner.pyx":425 + /* "pooling_inner.pyx":439 * * if not model.is_ft: * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< * * eff_sentences, eff_words = populate_base_s2v_config( */ - __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 425, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 439, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":429 + /* "pooling_inner.pyx":443 * eff_sentences, eff_words = populate_base_s2v_config( * &w2v, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 429, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 443, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 429, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_5); + __pyx_t_6 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 443, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":427 + /* "pooling_inner.pyx":441 * init_base_s2v_config(&w2v, model, target, memory) * * eff_sentences, eff_words = populate_base_s2v_config( # <<<<<<<<<<<<<< * &w2v, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 427, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_6, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 441, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; + __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { PyObject* sequence = __pyx_t_1; Py_ssize_t size = __Pyx_PySequence_SIZE(sequence); if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 427, __pyx_L1_error) + __PYX_ERR(0, 441, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { - __pyx_t_5 = PyTuple_GET_ITEM(sequence, 0); - __pyx_t_6 = PyTuple_GET_ITEM(sequence, 1); + __pyx_t_6 = PyTuple_GET_ITEM(sequence, 0); + __pyx_t_7 = PyTuple_GET_ITEM(sequence, 1); } else { - __pyx_t_5 = PyList_GET_ITEM(sequence, 0); - __pyx_t_6 = PyList_GET_ITEM(sequence, 1); + __pyx_t_6 = PyList_GET_ITEM(sequence, 0); + __pyx_t_7 = PyList_GET_ITEM(sequence, 1); } - __Pyx_INCREF(__pyx_t_5); __Pyx_INCREF(__pyx_t_6); + __Pyx_INCREF(__pyx_t_7); #else - __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 427, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_5); - __pyx_t_6 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 427, __pyx_L1_error) + __pyx_t_6 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 441, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); + __pyx_t_7 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 441, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_7); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 427, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_7); + __pyx_t_8 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_8)) __PYX_ERR(0, 441, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_8); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_8 = Py_TYPE(__pyx_t_7)->tp_iternext; - index = 0; __pyx_t_5 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_5)) goto __pyx_L4_unpacking_failed; - __Pyx_GOTREF(__pyx_t_5); - index = 1; __pyx_t_6 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_6)) goto __pyx_L4_unpacking_failed; + __pyx_t_9 = Py_TYPE(__pyx_t_8)->tp_iternext; + index = 0; __pyx_t_6 = __pyx_t_9(__pyx_t_8); if (unlikely(!__pyx_t_6)) goto __pyx_L4_unpacking_failed; __Pyx_GOTREF(__pyx_t_6); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 427, __pyx_L1_error) - __pyx_t_8 = NULL; - __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; + index = 1; __pyx_t_7 = __pyx_t_9(__pyx_t_8); if (unlikely(!__pyx_t_7)) goto __pyx_L4_unpacking_failed; + __Pyx_GOTREF(__pyx_t_7); + if (__Pyx_IternextUnpackEndCheck(__pyx_t_9(__pyx_t_8), 2) < 0) __PYX_ERR(0, 441, __pyx_L1_error) + __pyx_t_9 = NULL; + __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; goto __pyx_L5_unpacking_done; __pyx_L4_unpacking_failed:; - __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; - __pyx_t_8 = NULL; + __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; + __pyx_t_9 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 427, __pyx_L1_error) + __PYX_ERR(0, 441, __pyx_L1_error) __pyx_L5_unpacking_done:; } - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 427, __pyx_L1_error) - __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 427, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 441, __pyx_L1_error) __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_7); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 441, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; __pyx_v_eff_sentences = __pyx_t_2; - __pyx_v_eff_words = __pyx_t_9; + __pyx_v_eff_words = __pyx_t_10; - /* "pooling_inner.pyx":433 + /* "pooling_inner.pyx":447 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: * compute_base_sentence_pooling( */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 433, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 447, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 433, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 447, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_3 = ((!__pyx_t_4) != 0); - if (__pyx_t_3) { + __pyx_t_4 = ((!__pyx_t_5) != 0); + if (__pyx_t_4) { - /* "pooling_inner.pyx":434 + /* "pooling_inner.pyx":448 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -3273,7 +3348,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":435 + /* "pooling_inner.pyx":449 * if not model.hierarchical: * with nogil: * compute_base_sentence_pooling( # <<<<<<<<<<<<<< @@ -3283,7 +3358,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_base_sentence_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":434 + /* "pooling_inner.pyx":448 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -3302,7 +3377,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } } - /* "pooling_inner.pyx":433 + /* "pooling_inner.pyx":447 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< @@ -3312,7 +3387,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L6; } - /* "pooling_inner.pyx":440 + /* "pooling_inner.pyx":454 * ) * else: * with nogil: # <<<<<<<<<<<<<< @@ -3328,17 +3403,17 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":441 + /* "pooling_inner.pyx":455 * else: * with nogil: * compute_base_sentence_hier_pooling( # <<<<<<<<<<<<<< * &w2v, * eff_sentences, */ - __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences, __pyx_v_window_size); + __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences, __pyx_v_window_size, __pyx_v_window_stride); } - /* "pooling_inner.pyx":440 + /* "pooling_inner.pyx":454 * ) * else: * with nogil: # <<<<<<<<<<<<<< @@ -3359,7 +3434,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } __pyx_L6:; - /* "pooling_inner.pyx":424 + /* "pooling_inner.pyx":438 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< @@ -3369,7 +3444,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L3; } - /* "pooling_inner.pyx":447 + /* "pooling_inner.pyx":462 * ) * else: * init_ft_s2v_config(&ft, model, target, memory) # <<<<<<<<<<<<<< @@ -3377,101 +3452,101 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec * eff_sentences, eff_words = populate_ft_s2v_config( */ /*else*/ { - __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 447, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 462, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":451 + /* "pooling_inner.pyx":466 * eff_sentences, eff_words = populate_ft_s2v_config( * &ft, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 451, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 466, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_6 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 451, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_6); + __pyx_t_7 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 466, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_7); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":449 + /* "pooling_inner.pyx":464 * init_ft_s2v_config(&ft, model, target, memory) * * eff_sentences, eff_words = populate_ft_s2v_config( # <<<<<<<<<<<<<< * &ft, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_6, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 449, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_7, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 464, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; + __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { PyObject* sequence = __pyx_t_1; Py_ssize_t size = __Pyx_PySequence_SIZE(sequence); if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 449, __pyx_L1_error) + __PYX_ERR(0, 464, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { - __pyx_t_6 = PyTuple_GET_ITEM(sequence, 0); - __pyx_t_5 = PyTuple_GET_ITEM(sequence, 1); + __pyx_t_7 = PyTuple_GET_ITEM(sequence, 0); + __pyx_t_6 = PyTuple_GET_ITEM(sequence, 1); } else { - __pyx_t_6 = PyList_GET_ITEM(sequence, 0); - __pyx_t_5 = PyList_GET_ITEM(sequence, 1); + __pyx_t_7 = PyList_GET_ITEM(sequence, 0); + __pyx_t_6 = PyList_GET_ITEM(sequence, 1); } + __Pyx_INCREF(__pyx_t_7); __Pyx_INCREF(__pyx_t_6); - __Pyx_INCREF(__pyx_t_5); #else - __pyx_t_6 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 449, __pyx_L1_error) + __pyx_t_7 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 464, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_7); + __pyx_t_6 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 464, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); - __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 449, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_5); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_7 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 449, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_7); + __pyx_t_8 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_8)) __PYX_ERR(0, 464, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_8); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_8 = Py_TYPE(__pyx_t_7)->tp_iternext; - index = 0; __pyx_t_6 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_6)) goto __pyx_L13_unpacking_failed; + __pyx_t_9 = Py_TYPE(__pyx_t_8)->tp_iternext; + index = 0; __pyx_t_7 = __pyx_t_9(__pyx_t_8); if (unlikely(!__pyx_t_7)) goto __pyx_L13_unpacking_failed; + __Pyx_GOTREF(__pyx_t_7); + index = 1; __pyx_t_6 = __pyx_t_9(__pyx_t_8); if (unlikely(!__pyx_t_6)) goto __pyx_L13_unpacking_failed; __Pyx_GOTREF(__pyx_t_6); - index = 1; __pyx_t_5 = __pyx_t_8(__pyx_t_7); if (unlikely(!__pyx_t_5)) goto __pyx_L13_unpacking_failed; - __Pyx_GOTREF(__pyx_t_5); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_8(__pyx_t_7), 2) < 0) __PYX_ERR(0, 449, __pyx_L1_error) - __pyx_t_8 = NULL; - __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; + if (__Pyx_IternextUnpackEndCheck(__pyx_t_9(__pyx_t_8), 2) < 0) __PYX_ERR(0, 464, __pyx_L1_error) + __pyx_t_9 = NULL; + __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; goto __pyx_L14_unpacking_done; __pyx_L13_unpacking_failed:; - __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; - __pyx_t_8 = NULL; + __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; + __pyx_t_9 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 449, __pyx_L1_error) + __PYX_ERR(0, 464, __pyx_L1_error) __pyx_L14_unpacking_done:; } - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 449, __pyx_L1_error) + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_7); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 464, __pyx_L1_error) + __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 464, __pyx_L1_error) __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 449, __pyx_L1_error) - __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; - __pyx_v_eff_sentences = __pyx_t_9; + __pyx_v_eff_sentences = __pyx_t_10; __pyx_v_eff_words = __pyx_t_2; - /* "pooling_inner.pyx":455 + /* "pooling_inner.pyx":470 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: * compute_ft_sentence_pooling(&ft, eff_sentences) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 455, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 470, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_3 < 0)) __PYX_ERR(0, 455, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 470, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - __pyx_t_4 = ((!__pyx_t_3) != 0); - if (__pyx_t_4) { + __pyx_t_5 = ((!__pyx_t_4) != 0); + if (__pyx_t_5) { - /* "pooling_inner.pyx":456 + /* "pooling_inner.pyx":471 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -3486,7 +3561,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":457 + /* "pooling_inner.pyx":472 * if not model.hierarchical: * with nogil: * compute_ft_sentence_pooling(&ft, eff_sentences) # <<<<<<<<<<<<<< @@ -3496,7 +3571,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_ft_sentence_pooling((&__pyx_v_ft), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":456 + /* "pooling_inner.pyx":471 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -3515,7 +3590,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } } - /* "pooling_inner.pyx":455 + /* "pooling_inner.pyx":470 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< @@ -3525,7 +3600,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L15; } - /* "pooling_inner.pyx":459 + /* "pooling_inner.pyx":474 * compute_ft_sentence_pooling(&ft, eff_sentences) * else: * with nogil: # <<<<<<<<<<<<<< @@ -3541,17 +3616,17 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":460 + /* "pooling_inner.pyx":475 * else: * with nogil: * compute_ft_sentence_hier_pooling( # <<<<<<<<<<<<<< * &ft, * eff_sentences, */ - __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling((&__pyx_v_ft), __pyx_v_eff_sentences, __pyx_v_window_size); + __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling((&__pyx_v_ft), __pyx_v_eff_sentences, __pyx_v_window_size, __pyx_v_window_stride); } - /* "pooling_inner.pyx":459 + /* "pooling_inner.pyx":474 * compute_ft_sentence_pooling(&ft, eff_sentences) * else: * with nogil: # <<<<<<<<<<<<<< @@ -3574,7 +3649,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } __pyx_L3:; - /* "pooling_inner.pyx":466 + /* "pooling_inner.pyx":482 * ) * * return eff_sentences, eff_words # <<<<<<<<<<<<<< @@ -3582,23 +3657,23 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec * def init(): */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 466, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 482, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 466, __pyx_L1_error) - __Pyx_GOTREF(__pyx_t_5); - __pyx_t_6 = PyTuple_New(2); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 466, __pyx_L1_error) + __pyx_t_6 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 482, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); + __pyx_t_7 = PyTuple_New(2); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 482, __pyx_L1_error) + __Pyx_GOTREF(__pyx_t_7); __Pyx_GIVEREF(__pyx_t_1); - PyTuple_SET_ITEM(__pyx_t_6, 0, __pyx_t_1); - __Pyx_GIVEREF(__pyx_t_5); - PyTuple_SET_ITEM(__pyx_t_6, 1, __pyx_t_5); + PyTuple_SET_ITEM(__pyx_t_7, 0, __pyx_t_1); + __Pyx_GIVEREF(__pyx_t_6); + PyTuple_SET_ITEM(__pyx_t_7, 1, __pyx_t_6); __pyx_t_1 = 0; - __pyx_t_5 = 0; - __pyx_r = __pyx_t_6; __pyx_t_6 = 0; + __pyx_r = __pyx_t_7; + __pyx_t_7 = 0; goto __pyx_L0; - /* "pooling_inner.pyx":389 + /* "pooling_inner.pyx":402 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< @@ -3609,9 +3684,9 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec /* function exit code */ __pyx_L1_error:; __Pyx_XDECREF(__pyx_t_1); - __Pyx_XDECREF(__pyx_t_5); __Pyx_XDECREF(__pyx_t_6); __Pyx_XDECREF(__pyx_t_7); + __Pyx_XDECREF(__pyx_t_8); __Pyx_AddTraceback("pooling_inner.train_pooling_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __pyx_r = NULL; __pyx_L0:; @@ -3620,7 +3695,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec return __pyx_r; } -/* "pooling_inner.pyx":468 +/* "pooling_inner.pyx":484 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -3648,7 +3723,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("init", 0); - /* "pooling_inner.pyx":469 + /* "pooling_inner.pyx":485 * * def init(): * return 1 # <<<<<<<<<<<<<< @@ -3660,7 +3735,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __pyx_r = __pyx_int_1; goto __pyx_L0; - /* "pooling_inner.pyx":468 + /* "pooling_inner.pyx":484 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -6180,6 +6255,7 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = { {&__pyx_n_s_vocab, __pyx_k_vocab, sizeof(__pyx_k_vocab), 0, 0, 1, 1}, {&__pyx_n_s_w2v, __pyx_k_w2v, sizeof(__pyx_k_w2v), 0, 0, 1, 1}, {&__pyx_n_s_window_size, __pyx_k_window_size, sizeof(__pyx_k_window_size), 0, 0, 1, 1}, + {&__pyx_n_s_window_stride, __pyx_k_window_stride, sizeof(__pyx_k_window_stride), 0, 0, 1, 1}, {&__pyx_n_s_wv, __pyx_k_wv, sizeof(__pyx_k_wv), 0, 0, 1, 1}, {0, 0, 0, 0, 0, 0, 0} }; @@ -6274,26 +6350,26 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { __Pyx_GOTREF(__pyx_tuple__7); __Pyx_GIVEREF(__pyx_tuple__7); - /* "pooling_inner.pyx":389 + /* "pooling_inner.pyx":402 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_tuple__9 = PyTuple_Pack(9, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_window_size, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 389, __pyx_L1_error) + __pyx_tuple__9 = PyTuple_Pack(10, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_window_size, __pyx_n_s_window_stride, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 402, __pyx_L1_error) __Pyx_GOTREF(__pyx_tuple__9); __Pyx_GIVEREF(__pyx_tuple__9); - __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 9, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 389, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 389, __pyx_L1_error) + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 10, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 402, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 402, __pyx_L1_error) - /* "pooling_inner.pyx":468 + /* "pooling_inner.pyx":484 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 468, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 468, __pyx_L1_error) + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 484, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 484, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -6656,58 +6732,58 @@ if (!__Pyx_RefNanny) { if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_2) < 0) __PYX_ERR(0, 22, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":389 + /* "pooling_inner.pyx":402 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 389, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 402, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 389, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 402, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":468 + /* "pooling_inner.pyx":484 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 468, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 484, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 468, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 484, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":471 + /* "pooling_inner.pyx":487 * return 1 * * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 471, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 487, __pyx_L1_error) - /* "pooling_inner.pyx":472 + /* "pooling_inner.pyx":488 * * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 472, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 488, __pyx_L1_error) - /* "pooling_inner.pyx":473 + /* "pooling_inner.pyx":489 * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() # <<<<<<<<<<<<<< */ - __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 473, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 489, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 473, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 489, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 473, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 489, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; /* "pooling_inner.pyx":1 diff --git a/fse/models/pooling_inner.pyx b/fse/models/pooling_inner.pyx index 30be333..e8d2ea3 100644 --- a/fse/models/pooling_inner.pyx +++ b/fse/models/pooling_inner.pyx @@ -117,6 +117,7 @@ cdef void compute_base_sentence_hier_pooling( BaseSentenceVecsConfig *c, uINT_t num_sentences, uINT_t window_size, + REAL_t window_stride, ) nogil: """Perform optimized sentence-level hierarchical max pooling for BaseAny2Vec model. @@ -128,13 +129,14 @@ cdef void compute_base_sentence_hier_pooling( The number of sentences used to train the model. window_size : uINT_t The local window size. - + window_stride : REAL_t + The local window stride. + Notes ----- This routine does not provide oov support. """ - cdef: int size = c.size @@ -151,6 +153,10 @@ cdef void compute_base_sentence_hier_pooling( for sent_pos in range(sent_start, sent_end): sent_len += ONEF + + if (sent_len-ONEF) % window_stride != ZEROF: + continue + sent_row = c.sent_adresses[sent_pos] * size if sent_pos + window_size > sent_end: @@ -275,6 +281,7 @@ cdef void compute_ft_sentence_hier_pooling( FTSentenceVecsConfig *c, uINT_t num_sentences, uINT_t window_size, + REAL_t window_stride, ) nogil: """Perform optimized sentence-level hierarchical max pooling for FastText model. @@ -286,7 +293,9 @@ cdef void compute_ft_sentence_hier_pooling( The number of sentences used to train the model. window_size : uINT_t The local window size. - + window_stride : REAL_t + The local window stride. + Notes ----- This routine DOES provide oov support. @@ -314,6 +323,10 @@ cdef void compute_ft_sentence_hier_pooling( for sent_pos in range(sent_start, sent_end): sent_len += ONEF + + if (sent_len-ONEF) % window_stride != ZEROF: + continue + sent_row = c.sent_adresses[sent_pos] * size if sent_pos + window_size > sent_end: @@ -418,6 +431,7 @@ def train_pooling_cy( cdef uINT_t eff_sentences = 0 cdef uINT_t eff_words = 0 cdef uINT_t window_size = model.window_size + cdef REAL_t window_stride = model.window_stride cdef BaseSentenceVecsConfig w2v cdef FTSentenceVecsConfig ft @@ -442,6 +456,7 @@ def train_pooling_cy( &w2v, eff_sentences, window_size, + window_stride, ) else: init_ft_s2v_config(&ft, model, target, memory) @@ -461,6 +476,7 @@ def train_pooling_cy( &ft, eff_sentences, window_size, + window_stride, ) return eff_sentences, eff_words diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index 744210b..a94e3f3 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -85,6 +85,14 @@ def test_check_parameter_sanity(self): with self.assertRaises(ValueError): se._check_parameter_sanity() + se = MaxPooling(W2V, window_size=3, window_stride=0) + with self.assertRaises(ValueError): + se._check_parameter_sanity() + + se = MaxPooling(W2V, window_size=3, window_stride=4) + with self.assertRaises(ValueError): + se._check_parameter_sanity() + def test_train(self): self.assertEqual( (100, 1450), self.model.train([(s, i) for i, s in enumerate(SENTENCES)]) @@ -105,6 +113,8 @@ def test_do_train_job(self): ) self.assertEqual((105, DIM), self.model.sv.vectors.shape) + ### Basic Pooling Tests start here + def test_pool_train_np_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() @@ -133,6 +143,46 @@ def test_pool_train_cy_w2v(self): self.assertTrue((306 == self.model.sv[1]).all()) self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) + def test_pool_train_np_ft(self): + m = MaxPooling(FT) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + mem = m._get_thread_working_mem() + + output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + self.assertTrue((241 == m.sv[0]).all()) + self.assertTrue( + np.allclose(737413.9, m.sv[2]) + ) + self.assertTrue( + np.allclose(1080970.2, m.sv[3]) + ) + + def test_pool_train_cy_ft(self): + m = MaxPooling(FT) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + mem = m._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + self.assertTrue((241 == m.sv[0]).all()) + self.assertTrue( + np.allclose(737413.9, m.sv[2]) + ) + self.assertTrue( + np.allclose(1080970.2, m.sv[3]) + ) + def test_pool_cy_equal_np_w2v(self): m1 = MaxPooling(W2V) m1.prep.prepare_vectors( @@ -182,46 +232,6 @@ def test_pool_cy_equal_np_w2v_random(self): self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - def test_pool_train_np_ft(self): - m = MaxPooling(FT) - m.prep.prepare_vectors( - sv=m.sv, total_sentences=len(self.sentences), update=False - ) - m._pre_train_calls() - mem = m._get_thread_working_mem() - - output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) - - self.assertEqual((5, 19), output) - self.assertTrue((241 == m.sv[0]).all()) - self.assertTrue( - np.allclose(737413.9, m.sv[2]) - ) - self.assertTrue( - np.allclose(1080970.2, m.sv[3]) - ) - - def test_pool_train_cy_ft(self): - m = MaxPooling(FT) - m.prep.prepare_vectors( - sv=m.sv, total_sentences=len(self.sentences), update=False - ) - m._pre_train_calls() - mem = m._get_thread_working_mem() - - from fse.models.pooling_inner import train_pooling_cy - - output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) - - self.assertEqual((5, 19), output) - self.assertTrue((241 == m.sv[0]).all()) - self.assertTrue( - np.allclose(737413.9, m.sv[2]) - ) - self.assertTrue( - np.allclose(1080970.2, m.sv[3]) - ) - def test_pool_cy_equal_np_ft_random(self): ft = FastText(size=20, min_count=1) ft.build_vocab(SENTENCES) @@ -264,7 +274,7 @@ def test_pool_np_ft_non_negative(self): ### Hierarchical Tests start here - def test_hier_pooling_train_np_w2v(self): + def test_hier_pool_train_np_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() @@ -279,7 +289,7 @@ def test_hier_pooling_train_np_w2v(self): self.assertTrue((183 == self.model.sv[0]).all()) self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) - def test_hier_pooling_train_cy_w2v(self): + def test_hier_pool_train_cy_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() @@ -296,7 +306,7 @@ def test_hier_pooling_train_cy_w2v(self): self.assertTrue((183 == self.model.sv[0]).all()) self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) - def test_hier_pooling_train_np_ft(self): + def test_hier_pool_train_np_ft(self): m = MaxPooling(FT, hierarchical=True) m.prep.prepare_vectors( sv=m.sv, total_sentences=len(self.sentences), update=False @@ -318,7 +328,7 @@ def test_hier_pooling_train_np_ft(self): TODO: This unittest is thus a bit flawed. Maybe fix? """ - def test_hier_pooling_train_cy_ft(self): + def test_hier_pool_train_cy_ft(self): m = MaxPooling(FT, hierarchical=True) m.prep.prepare_vectors( sv=m.sv, total_sentences=len(self.sentences), update=False @@ -361,6 +371,36 @@ def test_hier_pool_cy_equal_np_w2v_random(self): self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + def test_hier_pool_cy_equal_np_ft_random(self): + ft = FastText(size=20, min_count=1) + ft.build_vocab(SENTENCES) + + m1 = MaxPooling(ft, hierarchical=True) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + + from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH + + m1.batch_ngrams = MAX_NGRAMS_IN_BATCH + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) + + m2 = MaxPooling(ft, hierarchical=True) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) + + self.assertEqual(o1, o2) + self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + def test_hier_pool_cy_w2v_non_negative(self): mpool = MaxPooling(W2V_R, hierarchical=True) mpool.train(self.sentences) @@ -371,6 +411,141 @@ def test_hier_pool_cy_ft_non_negative(self): mpool.train(self.sentences) self.assertTrue((mpool.sv.vectors >= 0).all()) + ### Hierarchical Test + Stride start here + + def test_hier_pool_stride_train_np_w2v(self): + self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + mem = self.model._get_thread_working_mem() + + self.model.hierarchical = True + self.model.window_stride = 5 + + output = train_pooling_np( + self.model, self.sentences, self.model.sv.vectors, mem + ) + self.model.hierarchical = False + self.model.window_stride = 1 + + self.assertEqual((5, 14), output) + self.assertTrue((183 == self.model.sv[0]).all()) + self.assertTrue((231 == self.model.sv[4]).all()) + + def test_hier_pool_stride_train_cy_w2v(self): + self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + mem = self.model._get_thread_working_mem() + + self.model.hierarchical = True + self.model.window_stride = 5 + + from fse.models.pooling_inner import train_pooling_cy + + output = train_pooling_cy( + self.model, self.sentences, self.model.sv.vectors, mem + ) + self.model.hierarchical = False + self.model.window_stride = 1 + + self.assertEqual((5, 14), output) + self.assertTrue((183 == self.model.sv[0]).all()) + self.assertTrue((231 == self.model.sv[4]).all()) + + def test_hier_pool_stride_train_np_ft(self): + m = MaxPooling(FT, hierarchical=True, window_stride=3) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + mem = m._get_thread_working_mem() + + output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + self.assertTrue((183 == m.sv[0]).all()) + self.assertTrue(np.allclose(368871.94, m.sv[2])) + self.assertTrue(np.allclose(961940.2, m.sv[3])) + + def test_hier_pool_stride_train_cy_ft(self): + m = MaxPooling(FT, hierarchical=True, window_stride=3) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + mem = m._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + self.assertTrue((183 == m.sv[0]).all()) + self.assertTrue(np.allclose(368871.94, m.sv[2])) + self.assertTrue(np.allclose(961940.2, m.sv[3])) + + def test_hier_pool_stride_cy_equal_np_w2v_random(self): + w2v = Word2Vec(min_count=1, size=DIM) + # Random initialization + w2v.build_vocab(SENTENCES) + + m1 = MaxPooling(w2v, hierarchical=True, window_stride=4) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + + m2 = MaxPooling(w2v, hierarchical=True, window_stride=4) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + + self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + def test_hier_pool_stride_cy_equal_np_ft_random(self): + ft = FastText(size=20, min_count=1) + ft.build_vocab(SENTENCES) + + m1 = MaxPooling(ft, hierarchical=True, window_stride=5) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + + from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH + + m1.batch_ngrams = MAX_NGRAMS_IN_BATCH + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) + + m2 = MaxPooling(ft, hierarchical=True, window_stride=5) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) + + self.assertEqual(o1, o2) + self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + def test_hier_pool_stride_np_w2v_non_negative(self): + mpool = MaxPooling(W2V_R, hierarchical=True, window_stride=4) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) + + def test_hier_pool_stride_np_ft_non_negative(self): + mpool = MaxPooling(FT_R, hierarchical=True, window_stride=4) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) if __name__ == "__main__": logging.basicConfig( From c864bf3cdaaf99594193f5e003207146990b9c38 Mon Sep 17 00:00:00 2001 From: OliverB Date: Mon, 2 Mar 2020 19:53:44 +0100 Subject: [PATCH 36/63] Updated todo --- fse/models/base_s2v.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index 145bcea..50690eb 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -143,7 +143,7 @@ def __init__( [ ] Stride Parameter: https://adeshpande3.github.io/A-Beginner%27s-Guide-To-Understanding-Convolutional-Neural-Networks-Part-2/ [X] Max Pooling [X] Hierarchical Max Pooling - [ ] Stride Paramter: https://adeshpande3.github.io/A-Beginner%27s-Guide-To-Understanding-Convolutional-Neural-Networks-Part-2/ + [X] Stride Paramter: https://adeshpande3.github.io/A-Beginner%27s-Guide-To-Understanding-Convolutional-Neural-Networks-Part-2/ [ ] discrete cosine transform -> somewhat easy, questionable [ ] valve -> unclear, not cited enough From 5ad3a48824aa93f970d8b3e9abbbf2d6893fd56d Mon Sep 17 00:00:00 2001 From: OliverB Date: Mon, 2 Mar 2020 19:55:44 +0100 Subject: [PATCH 37/63] Minor --- fse/test/test_pooling.py | 3 +-- 1 file changed, 1 insertion(+), 2 deletions(-) diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index a94e3f3..1ed85d4 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -45,7 +45,6 @@ FT_R = FastText(min_count=1, size=DIM) FT_R.build_vocab(SENTENCES) - class TestPoolingFunctions(unittest.TestCase): def setUp(self): @@ -541,7 +540,7 @@ def test_hier_pool_stride_np_w2v_non_negative(self): mpool = MaxPooling(W2V_R, hierarchical=True, window_stride=4) mpool.train(self.sentences) self.assertTrue((mpool.sv.vectors >= 0).all()) - + def test_hier_pool_stride_np_ft_non_negative(self): mpool = MaxPooling(FT_R, hierarchical=True, window_stride=4) mpool.train(self.sentences) From 565ae0e4ceb8604c3c1e17c560f9b4f83ec6fdfc Mon Sep 17 00:00:00 2001 From: OliverB Date: Mon, 2 Mar 2020 19:57:50 +0100 Subject: [PATCH 38/63] Updated Comments --- fse/models/average.py | 2 +- fse/models/average_inner.pxd | 2 +- fse/models/average_inner.pyx | 2 +- fse/models/base_s2v.py | 2 +- fse/models/pooling.py | 2 +- fse/models/pooling_inner.pyx | 2 +- fse/models/sentencevectors.py | 2 +- fse/models/sif.py | 2 +- fse/models/usif.py | 2 +- fse/models/utils.py | 2 +- fse/test/test_average.py | 2 +- fse/test/test_base_s2v.py | 2 +- fse/test/test_inputs.py | 2 +- fse/test/test_pooling.py | 2 +- fse/test/test_sentencevectors.py | 2 +- 15 files changed, 15 insertions(+), 15 deletions(-) diff --git a/fse/models/average.py b/fse/models/average.py index b068dde..21640a4 100644 --- a/fse/models/average.py +++ b/fse/models/average.py @@ -3,7 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. """This module implements the base class to compute average representations for sentences, using highly optimized C routines, data streaming and Pythonic interfaces. diff --git a/fse/models/average_inner.pxd b/fse/models/average_inner.pxd index 2b17205..2a1ac18 100644 --- a/fse/models/average_inner.pxd +++ b/fse/models/average_inner.pxd @@ -6,7 +6,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. cimport numpy as np diff --git a/fse/models/average_inner.pyx b/fse/models/average_inner.pyx index de57ecc..07e2b1d 100644 --- a/fse/models/average_inner.pyx +++ b/fse/models/average_inner.pyx @@ -7,7 +7,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. """Optimized cython functions for computing sentence embeddings""" diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index 50690eb..4444efa 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -3,7 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. """Base class containing common methods for training, using & evaluating sentence embeddings. A lot of the code is based on Gensim. I have to thank Radim Rehurek and the whole team diff --git a/fse/models/pooling.py b/fse/models/pooling.py index 2a1ebde..c1e6a74 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -3,7 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. """This module implements the base class to compute Max Pooling representations for sentences, using highly optimized C routines, data streaming and Pythonic interfaces. diff --git a/fse/models/pooling_inner.pyx b/fse/models/pooling_inner.pyx index e8d2ea3..2c74f90 100644 --- a/fse/models/pooling_inner.pyx +++ b/fse/models/pooling_inner.pyx @@ -7,7 +7,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. """Optimized cython functions for computing sentence embeddings""" diff --git a/fse/models/sentencevectors.py b/fse/models/sentencevectors.py index 6a3ac88..7590708 100644 --- a/fse/models/sentencevectors.py +++ b/fse/models/sentencevectors.py @@ -3,7 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. from __future__ import division diff --git a/fse/models/sif.py b/fse/models/sif.py index 6eeca2f..c79eb8b 100644 --- a/fse/models/sif.py +++ b/fse/models/sif.py @@ -3,7 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. from fse.models.average import Average from fse.models.utils import compute_principal_components, remove_principal_components diff --git a/fse/models/usif.py b/fse/models/usif.py index 29d6321..df1a74b 100644 --- a/fse/models/usif.py +++ b/fse/models/usif.py @@ -3,7 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. from fse.models.average import Average from fse.models.utils import compute_principal_components, remove_principal_components diff --git a/fse/models/utils.py b/fse/models/utils.py index aebab8a..32fc826 100644 --- a/fse/models/utils.py +++ b/fse/models/utils.py @@ -3,7 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. from sklearn.decomposition import TruncatedSVD diff --git a/fse/test/test_average.py b/fse/test/test_average.py index 27f066e..33f4090 100644 --- a/fse/test/test_average.py +++ b/fse/test/test_average.py @@ -3,7 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. """ Automated tests for checking the average model. diff --git a/fse/test/test_base_s2v.py b/fse/test/test_base_s2v.py index 818965b..126ab8a 100644 --- a/fse/test/test_base_s2v.py +++ b/fse/test/test_base_s2v.py @@ -3,7 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. """ diff --git a/fse/test/test_inputs.py b/fse/test/test_inputs.py index 8c050e5..ee6b4de 100644 --- a/fse/test/test_inputs.py +++ b/fse/test/test_inputs.py @@ -3,7 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. """ diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index 1ed85d4..2001665 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -3,7 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. """ Automated tests for checking the model. diff --git a/fse/test/test_sentencevectors.py b/fse/test/test_sentencevectors.py index 4a7680a..b740558 100644 --- a/fse/test/test_sentencevectors.py +++ b/fse/test/test_sentencevectors.py @@ -3,7 +3,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -# Licensed under GNU General Public License v3.0 +# For License information, see corresponding LICENSE file. """ From 08b7db32633aae0d18ba70f508bb5aecc6970f4f Mon Sep 17 00:00:00 2001 From: OliverB Date: Mon, 2 Mar 2020 20:12:19 +0100 Subject: [PATCH 39/63] Updated readme --- README.md | 19 +++++++++++-------- fse/models/pooling_inner.pyx | 6 ++++-- 2 files changed, 15 insertions(+), 10 deletions(-) diff --git a/README.md b/README.md index 05e8480..5035c99 100644 --- a/README.md +++ b/README.md @@ -20,14 +20,16 @@ Find the corresponding blog post(s) here: - [Visualizing 100,000 Amazon Products](https://towardsdatascience.com/vis-amz-83dea6fcb059) - [Sentence Embeddings. Fast, please!](https://towardsdatascience.com/fse-2b1ffa791cf9) -**fse** implements five algorithms for sentence embeddings. You can choose -between *unweighted sentence averages*, *smooth inverse frequency averages*, *unsupervised smooth inverse frequency averages*, *max pooling*, and *hierarchical max pooling*. +**fse** implements multiple algorithms for sentence embeddings. You can choose +between *unweighted sentence averages*, *smooth inverse frequency averages*, *unsupervised smooth inverse frequency averages*, and *max pooling*. All models support hierarchical estimation, similar to convolutional filters in CNNs. Key features of **fse** are: **[X]** Up to 500.000 sentences / second (1) -**[X]** Supports Average, SIF, uSIF, MaxPooling, and Hierarchical MaxPooling Embeddings +**[X]** Supports Average, SIF, uSIF, and MaxPooling Embeddings + +**[X]** All models can be estimated as hierarchical models (with window size and stride) **[X]** Full support for Gensims Word2Vec and all other compatible classes @@ -123,8 +125,6 @@ The models presented are based on - Unsupervised smooth inverse frequency embeddings [3] - MaxPooling / Hierarchical MaxPooling [5] -Credits to Radim Řehůřek and all contributors for the **awesome** library -and code that [Gensim](https://github.com/RaRe-Technologies/gensim) provides. A whole lot of the code found in this lib is based on Gensim. Results ------------ @@ -153,15 +153,17 @@ Changelog ------------- 0.1.16 from 0.1.15: -- Added MaxPooling / Hierarchical MaxPooling -- Added features to Sentencevectors +- Added Hierarchical (Convolutional) Embeddings for all Models +- Added MaxPooling +- Added Features to Sentencevectors - Added further unittests - Workaround for Numpy memmap issue (https://github.com/numpy/numpy/issues/13172) - Bugfixes for python 3.8 builds - Code refactoring to black style - SVD ram subsampling for SIF / uSIF (customizable, standard is 1 GB of RAM) - Minor fixes for nan-handling -- Minor fix for sentencevectors class +- Minor fixes for sentencevectors class +- Changed License 0.1.15 from 0.1.11: - Fixed major FT Ngram computation bug @@ -198,6 +200,7 @@ Author: Oliver Borchers Copyright (C) 2020 Oliver Borchers + Citation ------------- diff --git a/fse/models/pooling_inner.pyx b/fse/models/pooling_inner.pyx index 2c74f90..f01bb66 100644 --- a/fse/models/pooling_inner.pyx +++ b/fse/models/pooling_inner.pyx @@ -301,7 +301,6 @@ cdef void compute_ft_sentence_hier_pooling( This routine DOES provide oov support. """ - # The naming of the i,j,k vars is a bit different here cdef: int size = c.size @@ -469,7 +468,10 @@ def train_pooling_cy( if not model.hierarchical: with nogil: - compute_ft_sentence_pooling(&ft, eff_sentences) + compute_ft_sentence_pooling( + &ft, + eff_sentences + ) else: with nogil: compute_ft_sentence_hier_pooling( From 25d95b05bc6b32629a5aea908e74697ee6edaec1 Mon Sep 17 00:00:00 2001 From: OliverB Date: Tue, 3 Mar 2020 19:08:02 +0100 Subject: [PATCH 40/63] Memory fix --- fse/models/average_inner.c | 290 +++++++++++++++++------------------ fse/models/average_inner.pyx | 14 +- fse/models/base_s2v.py | 4 +- fse/models/pooling_inner.c | 290 +++++++++++++++++------------------ fse/test/test_base_s2v.py | 2 +- 5 files changed, 303 insertions(+), 297 deletions(-) diff --git a/fse/models/average_inner.c b/fse/models/average_inner.c index 7f309c5..20349c1 100644 --- a/fse/models/average_inner.c +++ b/fse/models/average_inner.c @@ -2008,7 +2008,7 @@ static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13a * c[0].size = model.sv.vector_size * * c[0].mem = (np.PyArray_DATA(memory[0])) # <<<<<<<<<<<<<< - * c[0].mem2 = (np.PyArray_DATA(memory[2])) + * c[0].mem2 = (np.PyArray_DATA(memory[1])) * */ __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_memory, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 62, __pyx_L1_error) @@ -2020,18 +2020,18 @@ static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13a /* "average_inner.pyx":63 * * c[0].mem = (np.PyArray_DATA(memory[0])) - * c[0].mem2 = (np.PyArray_DATA(memory[2])) # <<<<<<<<<<<<<< + * c[0].mem2 = (np.PyArray_DATA(memory[1])) # <<<<<<<<<<<<<< * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors)) */ - __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 63, __pyx_L1_error) + __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 63, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); if (!(likely(((__pyx_t_3) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_3, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 63, __pyx_L1_error) (__pyx_v_c[0]).mem2 = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_3))); __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0; /* "average_inner.pyx":65 - * c[0].mem2 = (np.PyArray_DATA(memory[2])) + * c[0].mem2 = (np.PyArray_DATA(memory[1])) * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors)) # <<<<<<<<<<<<<< * c[0].word_weights = (np.PyArray_DATA(model.word_weights)) @@ -2225,7 +2225,7 @@ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13ave * c[0].oov_weight = np.max(model.word_weights) * * c[0].mem = (np.PyArray_DATA(memory[0])) # <<<<<<<<<<<<<< - * c[0].mem2 = (np.PyArray_DATA(memory[2])) + * c[0].mem2 = (np.PyArray_DATA(memory[1])) * */ __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 0, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 100, __pyx_L1_error) @@ -2237,24 +2237,24 @@ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13ave /* "average_inner.pyx":101 * * c[0].mem = (np.PyArray_DATA(memory[0])) - * c[0].mem2 = (np.PyArray_DATA(memory[2])) # <<<<<<<<<<<<<< + * c[0].mem2 = (np.PyArray_DATA(memory[1])) # <<<<<<<<<<<<<< * - * memory[1].fill(ZERO) # Reset the ngram storage before filling the struct + * memory[2].fill(ZERO) # Reset the ngram storage before filling the struct */ - __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 101, __pyx_L1_error) + __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 101, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 101, __pyx_L1_error) (__pyx_v_c[0]).mem2 = ((__pyx_t_13average_inner_REAL_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; /* "average_inner.pyx":103 - * c[0].mem2 = (np.PyArray_DATA(memory[2])) + * c[0].mem2 = (np.PyArray_DATA(memory[1])) * - * memory[1].fill(ZERO) # Reset the ngram storage before filling the struct # <<<<<<<<<<<<<< - * c[0].subwords_idx = (np.PyArray_DATA(memory[1])) + * memory[2].fill(ZERO) # Reset the ngram storage before filling the struct # <<<<<<<<<<<<<< + * c[0].subwords_idx = (np.PyArray_DATA(memory[2])) * */ - __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 103, __pyx_L1_error) + __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 103, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __pyx_t_3 = __Pyx_PyObject_GetAttrStr(__pyx_t_4, __pyx_n_s_fill); if (unlikely(!__pyx_t_3)) __PYX_ERR(0, 103, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_3); @@ -2281,19 +2281,19 @@ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13ave /* "average_inner.pyx":104 * - * memory[1].fill(ZERO) # Reset the ngram storage before filling the struct - * c[0].subwords_idx = (np.PyArray_DATA(memory[1])) # <<<<<<<<<<<<<< + * memory[2].fill(ZERO) # Reset the ngram storage before filling the struct + * c[0].subwords_idx = (np.PyArray_DATA(memory[2])) # <<<<<<<<<<<<<< * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) */ - __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 1, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 104, __pyx_L1_error) + __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_memory, 2, long, 1, __Pyx_PyInt_From_long, 0, 0, 0); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 104, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); if (!(likely(((__pyx_t_1) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_1, __pyx_ptype_5numpy_ndarray))))) __PYX_ERR(0, 104, __pyx_L1_error) (__pyx_v_c[0]).subwords_idx = ((__pyx_t_13average_inner_uINT_t *)PyArray_DATA(((PyArrayObject *)__pyx_t_1))); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; /* "average_inner.pyx":106 - * c[0].subwords_idx = (np.PyArray_DATA(memory[1])) + * c[0].subwords_idx = (np.PyArray_DATA(memory[2])) * * c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) # <<<<<<<<<<<<<< * c[0].ngram_vectors = (np.PyArray_DATA(model.wv.vectors_ngrams)) @@ -3408,7 +3408,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 * * return eff_sents, eff_words # <<<<<<<<<<<<<< * - * cdef void compute_base_sentence_averages( + * */ __Pyx_XDECREF(__pyx_r); __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sents); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 231, __pyx_L1_error) @@ -3461,8 +3461,8 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 return __pyx_r; } -/* "average_inner.pyx":233 - * return eff_sents, eff_words +/* "average_inner.pyx":237 + * * * cdef void compute_base_sentence_averages( # <<<<<<<<<<<<<< * BaseSentenceVecsConfig *c, @@ -3489,7 +3489,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_t_13average_inner_uINT_t __pyx_t_7; int __pyx_t_8; - /* "average_inner.pyx":252 + /* "average_inner.pyx":256 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -3499,7 +3499,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "average_inner.pyx":260 + /* "average_inner.pyx":264 * REAL_t sent_len, inv_count * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -3511,7 +3511,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) { __pyx_v_sent_idx = __pyx_t_4; - /* "average_inner.pyx":261 + /* "average_inner.pyx":265 * * for sent_idx in range(num_sentences): * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -3520,7 +3520,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "average_inner.pyx":263 + /* "average_inner.pyx":267 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -3529,7 +3529,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "average_inner.pyx":264 + /* "average_inner.pyx":268 * * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -3538,7 +3538,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "average_inner.pyx":265 + /* "average_inner.pyx":269 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -3547,7 +3547,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "average_inner.pyx":267 + /* "average_inner.pyx":271 * sent_len = ZEROF * * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -3559,7 +3559,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { __pyx_v_sent_pos = __pyx_t_7; - /* "average_inner.pyx":268 + /* "average_inner.pyx":272 * * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -3568,7 +3568,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "average_inner.pyx":269 + /* "average_inner.pyx":273 * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< @@ -3577,7 +3577,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); - /* "average_inner.pyx":270 + /* "average_inner.pyx":274 * sent_len += ONEF * sent_row = c.sent_adresses[sent_pos] * size * word_row = c.word_indices[sent_pos] * size # <<<<<<<<<<<<<< @@ -3586,18 +3586,18 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_sent_pos]) * __pyx_v_size); - /* "average_inner.pyx":271 + /* "average_inner.pyx":275 * sent_row = c.sent_adresses[sent_pos] * size * word_row = c.word_indices[sent_pos] * size * word_idx = c.word_indices[sent_pos] # <<<<<<<<<<<<<< * - * saxpy( + * # Core loop */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_sent_pos]); - /* "average_inner.pyx":273 - * word_idx = c.word_indices[sent_pos] + /* "average_inner.pyx":278 * + * # Core loop * saxpy( # <<<<<<<<<<<<<< * &size, * &c.word_weights[word_idx], @@ -3605,9 +3605,9 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); } - /* "average_inner.pyx":282 - * ) + /* "average_inner.pyx":288 * + * # Post-Core Transform * if sent_len > ZEROF: # <<<<<<<<<<<<<< * inv_count = ONEF / sent_len * # If we perform the a*x on memory, the computation is compatible with many-to-one mappings @@ -3615,8 +3615,8 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ __pyx_t_8 = ((__pyx_v_sent_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_8) { - /* "average_inner.pyx":283 - * + /* "average_inner.pyx":289 + * # Post-Core Transform * if sent_len > ZEROF: * inv_count = ONEF / sent_len # <<<<<<<<<<<<<< * # If we perform the a*x on memory, the computation is compatible with many-to-one mappings @@ -3624,7 +3624,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_sent_len); - /* "average_inner.pyx":286 + /* "average_inner.pyx":292 * # If we perform the a*x on memory, the computation is compatible with many-to-one mappings * # because it doesn't rescale the overall result * saxpy( # <<<<<<<<<<<<<< @@ -3633,9 +3633,9 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&__pyx_v_13average_inner_ONE)); - /* "average_inner.pyx":282 - * ) + /* "average_inner.pyx":288 * + * # Post-Core Transform * if sent_len > ZEROF: # <<<<<<<<<<<<<< * inv_count = ONEF / sent_len * # If we perform the a*x on memory, the computation is compatible with many-to-one mappings @@ -3643,8 +3643,8 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ } } - /* "average_inner.pyx":233 - * return eff_sents, eff_words + /* "average_inner.pyx":237 + * * * cdef void compute_base_sentence_averages( # <<<<<<<<<<<<<< * BaseSentenceVecsConfig *c, @@ -3654,7 +3654,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ /* function exit code */ } -/* "average_inner.pyx":295 +/* "average_inner.pyx":301 * ) * * cdef void compute_ft_sentence_averages( # <<<<<<<<<<<<<< @@ -3691,7 +3691,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_13average_inner_uINT_t __pyx_t_11; __pyx_t_13average_inner_uINT_t __pyx_t_12; - /* "average_inner.pyx":314 + /* "average_inner.pyx":320 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -3701,7 +3701,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "average_inner.pyx":324 + /* "average_inner.pyx":330 * REAL_t sent_len * REAL_t inv_count, inv_ngram * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< @@ -3711,7 +3711,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_2 = __pyx_v_c->oov_weight; __pyx_v_oov_weight = __pyx_t_2; - /* "average_inner.pyx":327 + /* "average_inner.pyx":333 * * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -3723,7 +3723,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { __pyx_v_sent_idx = __pyx_t_5; - /* "average_inner.pyx":328 + /* "average_inner.pyx":334 * * for sent_idx in range(num_sentences): * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -3732,7 +3732,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "average_inner.pyx":329 + /* "average_inner.pyx":335 * for sent_idx in range(num_sentences): * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -3741,7 +3741,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "average_inner.pyx":330 + /* "average_inner.pyx":336 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -3750,7 +3750,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "average_inner.pyx":331 + /* "average_inner.pyx":337 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -3759,7 +3759,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "average_inner.pyx":333 + /* "average_inner.pyx":339 * sent_len = ZEROF * * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -3771,7 +3771,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { __pyx_v_sent_pos = __pyx_t_8; - /* "average_inner.pyx":334 + /* "average_inner.pyx":340 * * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -3780,7 +3780,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "average_inner.pyx":335 + /* "average_inner.pyx":341 * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< @@ -3789,7 +3789,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); - /* "average_inner.pyx":337 + /* "average_inner.pyx":343 * sent_row = c.sent_adresses[sent_pos] * size * * word_idx = c.word_indices[sent_pos] # <<<<<<<<<<<<<< @@ -3798,7 +3798,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_sent_pos]); - /* "average_inner.pyx":338 + /* "average_inner.pyx":344 * * word_idx = c.word_indices[sent_pos] * ngrams = c.subwords_idx_len[sent_pos] # <<<<<<<<<<<<<< @@ -3807,7 +3807,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_sent_pos]); - /* "average_inner.pyx":340 + /* "average_inner.pyx":346 * ngrams = c.subwords_idx_len[sent_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -3817,7 +3817,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); if (__pyx_t_9) { - /* "average_inner.pyx":341 + /* "average_inner.pyx":347 * * if ngrams == 0: * word_row = c.word_indices[sent_pos] * size # <<<<<<<<<<<<<< @@ -3826,7 +3826,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_sent_pos]) * __pyx_v_size); - /* "average_inner.pyx":342 + /* "average_inner.pyx":348 * if ngrams == 0: * word_row = c.word_indices[sent_pos] * size * saxpy( # <<<<<<<<<<<<<< @@ -3835,7 +3835,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); - /* "average_inner.pyx":340 + /* "average_inner.pyx":346 * ngrams = c.subwords_idx_len[sent_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -3845,7 +3845,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ goto __pyx_L7; } - /* "average_inner.pyx":351 + /* "average_inner.pyx":357 * ) * else: * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< @@ -3855,7 +3855,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ /*else*/ { __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); - /* "average_inner.pyx":352 + /* "average_inner.pyx":358 * else: * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): # <<<<<<<<<<<<<< @@ -3867,7 +3867,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ for (__pyx_t_12 = 0; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { __pyx_v_ngram_pos = __pyx_t_12; - /* "average_inner.pyx":353 + /* "average_inner.pyx":359 * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size # <<<<<<<<<<<<<< @@ -3876,7 +3876,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_sent_pos * 40) + __pyx_v_ngram_pos)]) * __pyx_v_size); - /* "average_inner.pyx":354 + /* "average_inner.pyx":360 * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size * saxpy( # <<<<<<<<<<<<<< @@ -3889,7 +3889,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_L7:; } - /* "average_inner.pyx":363 + /* "average_inner.pyx":369 * ) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< @@ -3899,7 +3899,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ __pyx_t_9 = ((__pyx_v_sent_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_9) { - /* "average_inner.pyx":364 + /* "average_inner.pyx":370 * * if sent_len > ZEROF: * inv_count = ONEF / sent_len # <<<<<<<<<<<<<< @@ -3908,7 +3908,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_sent_len); - /* "average_inner.pyx":365 + /* "average_inner.pyx":371 * if sent_len > ZEROF: * inv_count = ONEF / sent_len * saxpy( # <<<<<<<<<<<<<< @@ -3917,7 +3917,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row])), (&__pyx_v_13average_inner_ONE)); - /* "average_inner.pyx":363 + /* "average_inner.pyx":369 * ) * * if sent_len > ZEROF: # <<<<<<<<<<<<<< @@ -3927,7 +3927,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ } } - /* "average_inner.pyx":295 + /* "average_inner.pyx":301 * ) * * cdef void compute_ft_sentence_averages( # <<<<<<<<<<<<<< @@ -3938,7 +3938,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ /* function exit code */ } -/* "average_inner.pyx":374 +/* "average_inner.pyx":380 * ) * * def train_average_cy( # <<<<<<<<<<<<<< @@ -3985,23 +3985,23 @@ static PyObject *__pyx_pw_13average_inner_1train_average_cy(PyObject *__pyx_self case 1: if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 1); __PYX_ERR(0, 374, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 1); __PYX_ERR(0, 380, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 2: if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 2); __PYX_ERR(0, 374, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 2); __PYX_ERR(0, 380, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 3: if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 3); __PYX_ERR(0, 374, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, 3); __PYX_ERR(0, 380, __pyx_L3_error) } } if (unlikely(kw_args > 0)) { - if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_average_cy") < 0)) __PYX_ERR(0, 374, __pyx_L3_error) + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_average_cy") < 0)) __PYX_ERR(0, 380, __pyx_L3_error) } } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { goto __pyx_L5_argtuple_error; @@ -4018,7 +4018,7 @@ static PyObject *__pyx_pw_13average_inner_1train_average_cy(PyObject *__pyx_self } goto __pyx_L4_argument_unpacking_done; __pyx_L5_argtuple_error:; - __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 374, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_average_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 380, __pyx_L3_error) __pyx_L3_error:; __Pyx_AddTraceback("average_inner.train_average_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __Pyx_RefNannyFinishContext(); @@ -4049,7 +4049,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_t_13average_inner_uINT_t __pyx_t_9; __Pyx_RefNannySetupContext("train_average_cy", 0); - /* "average_inner.pyx":403 + /* "average_inner.pyx":409 * """ * * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< @@ -4058,7 +4058,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_sentences = 0; - /* "average_inner.pyx":404 + /* "average_inner.pyx":410 * * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< @@ -4067,52 +4067,52 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_words = 0; - /* "average_inner.pyx":408 + /* "average_inner.pyx":414 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< * init_base_s2v_config(&w2v, model, target, memory) * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 408, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 414, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 408, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 414, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_3 = ((!__pyx_t_2) != 0); if (__pyx_t_3) { - /* "average_inner.pyx":409 + /* "average_inner.pyx":415 * * if not model.is_ft: * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< * * eff_sentences, eff_words = populate_base_s2v_config( */ - __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 409, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 415, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":413 + /* "average_inner.pyx":419 * eff_sentences, eff_words = populate_base_s2v_config( * &w2v, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences, * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 413, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 419, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 413, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 419, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":411 + /* "average_inner.pyx":417 * init_base_s2v_config(&w2v, model, target, memory) * * eff_sentences, eff_words = populate_base_s2v_config( # <<<<<<<<<<<<<< * &w2v, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 411, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 417, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -4121,7 +4121,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 411, __pyx_L1_error) + __PYX_ERR(0, 417, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -4134,15 +4134,15 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_4); __Pyx_INCREF(__pyx_t_5); #else - __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 411, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 417, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 411, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 417, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 411, __pyx_L1_error) + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 417, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; @@ -4150,7 +4150,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_4); index = 1; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L4_unpacking_failed; __Pyx_GOTREF(__pyx_t_5); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 411, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 417, __pyx_L1_error) __pyx_t_7 = NULL; __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; goto __pyx_L5_unpacking_done; @@ -4158,17 +4158,17 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_t_7 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 411, __pyx_L1_error) + __PYX_ERR(0, 417, __pyx_L1_error) __pyx_L5_unpacking_done:; } - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 411, __pyx_L1_error) + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 417, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 411, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 417, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; __pyx_v_eff_sentences = __pyx_t_8; __pyx_v_eff_words = __pyx_t_9; - /* "average_inner.pyx":417 + /* "average_inner.pyx":423 * ) * * with nogil: # <<<<<<<<<<<<<< @@ -4183,7 +4183,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "average_inner.pyx":418 + /* "average_inner.pyx":424 * * with nogil: * compute_base_sentence_averages( # <<<<<<<<<<<<<< @@ -4193,7 +4193,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_f_13average_inner_compute_base_sentence_averages((&__pyx_v_w2v), __pyx_v_eff_sentences); } - /* "average_inner.pyx":417 + /* "average_inner.pyx":423 * ) * * with nogil: # <<<<<<<<<<<<<< @@ -4212,7 +4212,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec } } - /* "average_inner.pyx":408 + /* "average_inner.pyx":414 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< @@ -4222,7 +4222,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec goto __pyx_L3; } - /* "average_inner.pyx":423 + /* "average_inner.pyx":429 * ) * else: * init_ft_s2v_config( # <<<<<<<<<<<<<< @@ -4231,38 +4231,38 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec */ /*else*/ { - /* "average_inner.pyx":427 + /* "average_inner.pyx":433 * model, * target, * memory, # <<<<<<<<<<<<<< * ) * */ - __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 423, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 429, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":432 + /* "average_inner.pyx":438 * eff_sentences, eff_words = populate_ft_s2v_config( * &ft, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences, * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 432, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 438, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 432, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 438, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":430 + /* "average_inner.pyx":436 * ) * * eff_sentences, eff_words = populate_ft_s2v_config( # <<<<<<<<<<<<<< * &ft, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 430, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 436, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -4271,7 +4271,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 430, __pyx_L1_error) + __PYX_ERR(0, 436, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -4284,15 +4284,15 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_5); __Pyx_INCREF(__pyx_t_4); #else - __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 430, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 436, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); - __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 430, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 436, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 430, __pyx_L1_error) + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 436, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; @@ -4300,7 +4300,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_5); index = 1; __pyx_t_4 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_4)) goto __pyx_L9_unpacking_failed; __Pyx_GOTREF(__pyx_t_4); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 430, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 436, __pyx_L1_error) __pyx_t_7 = NULL; __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; goto __pyx_L10_unpacking_done; @@ -4308,17 +4308,17 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_t_7 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 430, __pyx_L1_error) + __PYX_ERR(0, 436, __pyx_L1_error) __pyx_L10_unpacking_done:; } - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 430, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 436, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 430, __pyx_L1_error) + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 436, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; __pyx_v_eff_sentences = __pyx_t_9; __pyx_v_eff_words = __pyx_t_8; - /* "average_inner.pyx":436 + /* "average_inner.pyx":442 * ) * * with nogil: # <<<<<<<<<<<<<< @@ -4333,7 +4333,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "average_inner.pyx":437 + /* "average_inner.pyx":443 * * with nogil: * compute_ft_sentence_averages( # <<<<<<<<<<<<<< @@ -4343,7 +4343,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_f_13average_inner_compute_ft_sentence_averages((&__pyx_v_ft), __pyx_v_eff_sentences); } - /* "average_inner.pyx":436 + /* "average_inner.pyx":442 * ) * * with nogil: # <<<<<<<<<<<<<< @@ -4364,7 +4364,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec } __pyx_L3:; - /* "average_inner.pyx":442 + /* "average_inner.pyx":448 * ) * * return eff_sentences, eff_words # <<<<<<<<<<<<<< @@ -4372,11 +4372,11 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec * def init(): */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 442, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 448, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 442, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 448, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 442, __pyx_L1_error) + __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 448, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_1); @@ -4388,7 +4388,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __pyx_t_5 = 0; goto __pyx_L0; - /* "average_inner.pyx":374 + /* "average_inner.pyx":380 * ) * * def train_average_cy( # <<<<<<<<<<<<<< @@ -4410,7 +4410,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec return __pyx_r; } -/* "average_inner.pyx":444 +/* "average_inner.pyx":450 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -4438,7 +4438,7 @@ static PyObject *__pyx_pf_13average_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("init", 0); - /* "average_inner.pyx":445 + /* "average_inner.pyx":451 * * def init(): * return 1 # <<<<<<<<<<<<<< @@ -4450,7 +4450,7 @@ static PyObject *__pyx_pf_13average_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __pyx_r = __pyx_int_1; goto __pyx_L0; - /* "average_inner.pyx":444 + /* "average_inner.pyx":450 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -6998,7 +6998,7 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = { }; static CYTHON_SMALL_CODE int __Pyx_InitCachedBuiltins(void) { __pyx_builtin_enumerate = __Pyx_GetBuiltinName(__pyx_n_s_enumerate); if (!__pyx_builtin_enumerate) __PYX_ERR(0, 217, __pyx_L1_error) - __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 260, __pyx_L1_error) + __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 264, __pyx_L1_error) __pyx_builtin_ValueError = __Pyx_GetBuiltinName(__pyx_n_s_ValueError); if (!__pyx_builtin_ValueError) __PYX_ERR(1, 272, __pyx_L1_error) __pyx_builtin_RuntimeError = __Pyx_GetBuiltinName(__pyx_n_s_RuntimeError); if (!__pyx_builtin_RuntimeError) __PYX_ERR(1, 856, __pyx_L1_error) __pyx_builtin_ImportError = __Pyx_GetBuiltinName(__pyx_n_s_ImportError); if (!__pyx_builtin_ImportError) __PYX_ERR(1, 1038, __pyx_L1_error) @@ -7088,26 +7088,26 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { __Pyx_GOTREF(__pyx_tuple__7); __Pyx_GIVEREF(__pyx_tuple__7); - /* "average_inner.pyx":374 + /* "average_inner.pyx":380 * ) * * def train_average_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 374, __pyx_L1_error) + __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 380, __pyx_L1_error) __Pyx_GOTREF(__pyx_tuple__9); __Pyx_GIVEREF(__pyx_tuple__9); - __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_train_average_cy, 374, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 374, __pyx_L1_error) + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_train_average_cy, 380, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 380, __pyx_L1_error) - /* "average_inner.pyx":444 + /* "average_inner.pyx":450 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_init, 444, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 444, __pyx_L1_error) + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_init, 450, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 450, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -7563,58 +7563,58 @@ if (!__Pyx_RefNanny) { */ __pyx_v_13average_inner_ZEROF = ((__pyx_t_13average_inner_REAL_t)0.0); - /* "average_inner.pyx":374 + /* "average_inner.pyx":380 * ) * * def train_average_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_1train_average_cy, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 374, __pyx_L1_error) + __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_1train_average_cy, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 380, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_average_cy, __pyx_t_1) < 0) __PYX_ERR(0, 374, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_average_cy, __pyx_t_1) < 0) __PYX_ERR(0, 380, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":444 + /* "average_inner.pyx":450 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_3init, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 444, __pyx_L1_error) + __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_3init, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 450, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_1) < 0) __PYX_ERR(0, 444, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_1) < 0) __PYX_ERR(0, 450, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":447 + /* "average_inner.pyx":453 * return 1 * * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 447, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 453, __pyx_L1_error) - /* "average_inner.pyx":448 + /* "average_inner.pyx":454 * * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 448, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 454, __pyx_L1_error) - /* "average_inner.pyx":449 + /* "average_inner.pyx":455 * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() # <<<<<<<<<<<<<< */ - __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_init); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 449, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_init); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 455, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_CallNoArg(__pyx_t_1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 449, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_CallNoArg(__pyx_t_1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 455, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_2) < 0) __PYX_ERR(0, 449, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_2) < 0) __PYX_ERR(0, 455, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; /* "average_inner.pyx":1 diff --git a/fse/models/average_inner.pyx b/fse/models/average_inner.pyx index 07e2b1d..bc85f28 100644 --- a/fse/models/average_inner.pyx +++ b/fse/models/average_inner.pyx @@ -60,7 +60,7 @@ cdef init_base_s2v_config( c[0].size = model.sv.vector_size c[0].mem = (np.PyArray_DATA(memory[0])) - c[0].mem2 = (np.PyArray_DATA(memory[2])) + c[0].mem2 = (np.PyArray_DATA(memory[1])) c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors)) c[0].word_weights = (np.PyArray_DATA(model.word_weights)) @@ -98,10 +98,10 @@ cdef init_ft_s2v_config( c[0].oov_weight = np.max(model.word_weights) c[0].mem = (np.PyArray_DATA(memory[0])) - c[0].mem2 = (np.PyArray_DATA(memory[2])) + c[0].mem2 = (np.PyArray_DATA(memory[1])) - memory[1].fill(ZERO) # Reset the ngram storage before filling the struct - c[0].subwords_idx = (np.PyArray_DATA(memory[1])) + memory[2].fill(ZERO) # Reset the ngram storage before filling the struct + c[0].subwords_idx = (np.PyArray_DATA(memory[2])) c[0].word_vectors = (np.PyArray_DATA(model.wv.vectors_vocab)) c[0].ngram_vectors = (np.PyArray_DATA(model.wv.vectors_ngrams)) @@ -230,6 +230,10 @@ cdef object populate_ft_s2v_config( return eff_sents, eff_words + + + + cdef void compute_base_sentence_averages( BaseSentenceVecsConfig *c, uINT_t num_sentences, @@ -270,6 +274,7 @@ cdef void compute_base_sentence_averages( word_row = c.word_indices[sent_pos] * size word_idx = c.word_indices[sent_pos] + # Core loop saxpy( &size, &c.word_weights[word_idx], @@ -279,6 +284,7 @@ cdef void compute_base_sentence_averages( &ONE ) + # Post-Core Transform if sent_len > ZEROF: inv_count = ONEF / sent_len # If we perform the a*x on memory, the computation is compatible with many-to-one mappings diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index 4444efa..d92a187 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -541,13 +541,13 @@ def _get_thread_working_mem(self) -> tuple: """ mem = zeros_aligned(self.sv.vector_size, dtype=REAL) + mem2 = zeros_aligned(self.sv.vector_size, dtype=REAL) if self.is_ft: oov_mem = zeros_aligned((self.batch_words, self.batch_ngrams), dtype=uINT) else: # Save memory if not using FT oov_mem = 0 - mem2 = zeros_aligned(self.sv.vector_size, dtype=REAL) - return (mem, oov_mem, mem2) + return (mem, mem2, oov_mem) def _do_train_job( self, data_iterable: List[tuple], target: ndarray, memory: tuple diff --git a/fse/models/pooling_inner.c b/fse/models/pooling_inner.c index f8d4a14..e67daa0 100644 --- a/fse/models/pooling_inner.c +++ b/fse/models/pooling_inner.c @@ -2667,7 +2667,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_13average_inner_uINT_t __pyx_t_14; __pyx_t_13average_inner_uINT_t __pyx_t_15; - /* "pooling_inner.pyx":307 + /* "pooling_inner.pyx":306 * * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -2677,7 +2677,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":317 + /* "pooling_inner.pyx":316 * REAL_t sent_len, win_len * REAL_t inv_count, inv_ngram * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< @@ -2687,7 +2687,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_2 = __pyx_v_c->oov_weight; __pyx_v_oov_weight = __pyx_t_2; - /* "pooling_inner.pyx":319 + /* "pooling_inner.pyx":318 * REAL_t oov_weight = c.oov_weight * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -2699,7 +2699,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { __pyx_v_sent_idx = __pyx_t_5; - /* "pooling_inner.pyx":320 + /* "pooling_inner.pyx":319 * * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -2708,7 +2708,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":321 + /* "pooling_inner.pyx":320 * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -2717,7 +2717,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":322 + /* "pooling_inner.pyx":321 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -2726,7 +2726,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":324 + /* "pooling_inner.pyx":323 * sent_len = ZEROF * * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -2738,7 +2738,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { __pyx_v_sent_pos = __pyx_t_8; - /* "pooling_inner.pyx":325 + /* "pooling_inner.pyx":324 * * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -2747,7 +2747,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":327 + /* "pooling_inner.pyx":326 * sent_len += ONEF * * if (sent_len-ONEF) % window_stride != ZEROF: # <<<<<<<<<<<<<< @@ -2757,7 +2757,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = ((fmodf((__pyx_v_sent_len - __pyx_v_13average_inner_ONEF), __pyx_v_window_stride) != __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":328 + /* "pooling_inner.pyx":327 * * if (sent_len-ONEF) % window_stride != ZEROF: * continue # <<<<<<<<<<<<<< @@ -2766,7 +2766,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ goto __pyx_L5_continue; - /* "pooling_inner.pyx":327 + /* "pooling_inner.pyx":326 * sent_len += ONEF * * if (sent_len-ONEF) % window_stride != ZEROF: # <<<<<<<<<<<<<< @@ -2775,7 +2775,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ } - /* "pooling_inner.pyx":330 + /* "pooling_inner.pyx":329 * continue * * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< @@ -2784,7 +2784,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":332 + /* "pooling_inner.pyx":331 * sent_row = c.sent_adresses[sent_pos] * size * * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< @@ -2794,7 +2794,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = (((__pyx_v_sent_pos + __pyx_v_window_size) > __pyx_v_sent_end) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":333 + /* "pooling_inner.pyx":332 * * if sent_pos + window_size > sent_end: * window_end = sent_end # <<<<<<<<<<<<<< @@ -2803,7 +2803,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_window_end = __pyx_v_sent_end; - /* "pooling_inner.pyx":332 + /* "pooling_inner.pyx":331 * sent_row = c.sent_adresses[sent_pos] * size * * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< @@ -2813,7 +2813,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py goto __pyx_L8; } - /* "pooling_inner.pyx":335 + /* "pooling_inner.pyx":334 * window_end = sent_end * else: * window_end = sent_pos + window_size # <<<<<<<<<<<<<< @@ -2825,7 +2825,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py } __pyx_L8:; - /* "pooling_inner.pyx":338 + /* "pooling_inner.pyx":337 * * # Compute the locally averaged window * win_len = ZEROF # <<<<<<<<<<<<<< @@ -2834,7 +2834,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_win_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":339 + /* "pooling_inner.pyx":338 * # Compute the locally averaged window * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2843,7 +2843,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":340 + /* "pooling_inner.pyx":339 * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2852,7 +2852,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":341 + /* "pooling_inner.pyx":340 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * for window_pos in range(sent_pos, window_end): # <<<<<<<<<<<<<< @@ -2864,7 +2864,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_12 = __pyx_v_sent_pos; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { __pyx_v_window_pos = __pyx_t_12; - /* "pooling_inner.pyx":342 + /* "pooling_inner.pyx":341 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * for window_pos in range(sent_pos, window_end): * win_len += ONEF # <<<<<<<<<<<<<< @@ -2873,7 +2873,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_win_len = (__pyx_v_win_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":343 + /* "pooling_inner.pyx":342 * for window_pos in range(sent_pos, window_end): * win_len += ONEF * ngrams = c.subwords_idx_len[window_pos] # <<<<<<<<<<<<<< @@ -2882,7 +2882,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_window_pos]); - /* "pooling_inner.pyx":345 + /* "pooling_inner.pyx":344 * ngrams = c.subwords_idx_len[window_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2892,7 +2892,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":346 + /* "pooling_inner.pyx":345 * * if ngrams == 0: * word_row = c.word_indices[window_pos] * size # <<<<<<<<<<<<<< @@ -2901,7 +2901,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_window_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":347 + /* "pooling_inner.pyx":346 * if ngrams == 0: * word_row = c.word_indices[window_pos] * size * word_idx = c.word_indices[window_pos] # <<<<<<<<<<<<<< @@ -2910,7 +2910,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_window_pos]); - /* "pooling_inner.pyx":349 + /* "pooling_inner.pyx":348 * word_idx = c.word_indices[window_pos] * * saxpy( # <<<<<<<<<<<<<< @@ -2919,7 +2919,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); - /* "pooling_inner.pyx":345 + /* "pooling_inner.pyx":344 * ngrams = c.subwords_idx_len[window_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2929,7 +2929,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py goto __pyx_L11; } - /* "pooling_inner.pyx":359 + /* "pooling_inner.pyx":358 * * else: * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2939,7 +2939,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py /*else*/ { (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":360 + /* "pooling_inner.pyx":359 * else: * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< @@ -2948,7 +2948,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); - /* "pooling_inner.pyx":361 + /* "pooling_inner.pyx":360 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): # <<<<<<<<<<<<<< @@ -2960,7 +2960,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_15 = 0; __pyx_t_15 < __pyx_t_14; __pyx_t_15+=1) { __pyx_v_ngram_pos = __pyx_t_15; - /* "pooling_inner.pyx":362 + /* "pooling_inner.pyx":361 * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(window_pos * MAX_NGRAMS)+ngram_pos] * size # <<<<<<<<<<<<<< @@ -2969,7 +2969,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_window_pos * 40) + __pyx_v_ngram_pos)]) * __pyx_v_size); - /* "pooling_inner.pyx":363 + /* "pooling_inner.pyx":362 * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(window_pos * MAX_NGRAMS)+ngram_pos] * size * saxpy( # <<<<<<<<<<<<<< @@ -2979,7 +2979,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_ngram), (&(__pyx_v_c->ngram_vectors[__pyx_v_ngram_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE)); } - /* "pooling_inner.pyx":372 + /* "pooling_inner.pyx":371 * ) * * saxpy( # <<<<<<<<<<<<<< @@ -2991,7 +2991,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_L11:; } - /* "pooling_inner.pyx":381 + /* "pooling_inner.pyx":380 * ) * * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -3000,7 +3000,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":383 + /* "pooling_inner.pyx":382 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * # Rescale for dynamic window size * if win_len > ZEROF: # <<<<<<<<<<<<<< @@ -3010,7 +3010,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = ((__pyx_v_win_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":384 + /* "pooling_inner.pyx":383 * # Rescale for dynamic window size * if win_len > ZEROF: * inv_count = ONEF / win_len # <<<<<<<<<<<<<< @@ -3019,7 +3019,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_win_len); - /* "pooling_inner.pyx":385 + /* "pooling_inner.pyx":384 * if win_len > ZEROF: * inv_count = ONEF / win_len * saxpy( # <<<<<<<<<<<<<< @@ -3028,7 +3028,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE)); - /* "pooling_inner.pyx":383 + /* "pooling_inner.pyx":382 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * # Rescale for dynamic window size * if win_len > ZEROF: # <<<<<<<<<<<<<< @@ -3037,7 +3037,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ } - /* "pooling_inner.pyx":394 + /* "pooling_inner.pyx":393 * ) * * swrmax_pool( # <<<<<<<<<<<<<< @@ -3060,7 +3060,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py /* function exit code */ } -/* "pooling_inner.pyx":402 +/* "pooling_inner.pyx":401 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< @@ -3107,23 +3107,23 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self case 1: if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 402, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 401, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 2: if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 402, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 401, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 3: if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 402, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 401, __pyx_L3_error) } } if (unlikely(kw_args > 0)) { - if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 402, __pyx_L3_error) + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 401, __pyx_L3_error) } } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { goto __pyx_L5_argtuple_error; @@ -3140,7 +3140,7 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self } goto __pyx_L4_argument_unpacking_done; __pyx_L5_argtuple_error:; - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 402, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 401, __pyx_L3_error) __pyx_L3_error:; __Pyx_AddTraceback("pooling_inner.train_pooling_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __Pyx_RefNannyFinishContext(); @@ -3174,7 +3174,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_13average_inner_uINT_t __pyx_t_10; __Pyx_RefNannySetupContext("train_pooling_cy", 0); - /* "pooling_inner.pyx":431 + /* "pooling_inner.pyx":430 * """ * * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< @@ -3183,7 +3183,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_sentences = 0; - /* "pooling_inner.pyx":432 + /* "pooling_inner.pyx":431 * * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< @@ -3192,78 +3192,78 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_words = 0; - /* "pooling_inner.pyx":433 + /* "pooling_inner.pyx":432 * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 * cdef uINT_t window_size = model.window_size # <<<<<<<<<<<<<< * cdef REAL_t window_stride = model.window_stride * cdef BaseSentenceVecsConfig w2v */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_size); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 433, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_size); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 432, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_1); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 433, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_1); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 432, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_v_window_size = ((__pyx_t_13average_inner_uINT_t)__pyx_t_2); - /* "pooling_inner.pyx":434 + /* "pooling_inner.pyx":433 * cdef uINT_t eff_words = 0 * cdef uINT_t window_size = model.window_size * cdef REAL_t window_stride = model.window_stride # <<<<<<<<<<<<<< * cdef BaseSentenceVecsConfig w2v * cdef FTSentenceVecsConfig ft */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_stride); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 434, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_stride); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 433, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __pyx_PyFloat_AsFloat(__pyx_t_1); if (unlikely((__pyx_t_3 == ((npy_float32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 434, __pyx_L1_error) + __pyx_t_3 = __pyx_PyFloat_AsFloat(__pyx_t_1); if (unlikely((__pyx_t_3 == ((npy_float32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 433, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_v_window_stride = ((__pyx_t_13average_inner_REAL_t)__pyx_t_3); - /* "pooling_inner.pyx":438 + /* "pooling_inner.pyx":437 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< * init_base_s2v_config(&w2v, model, target, memory) * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 438, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 437, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 438, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 437, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_5 = ((!__pyx_t_4) != 0); if (__pyx_t_5) { - /* "pooling_inner.pyx":439 + /* "pooling_inner.pyx":438 * * if not model.is_ft: * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< * * eff_sentences, eff_words = populate_base_s2v_config( */ - __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 439, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 438, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":443 + /* "pooling_inner.pyx":442 * eff_sentences, eff_words = populate_base_s2v_config( * &w2v, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 443, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 442, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_6 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 443, __pyx_L1_error) + __pyx_t_6 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 442, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":441 + /* "pooling_inner.pyx":440 * init_base_s2v_config(&w2v, model, target, memory) * * eff_sentences, eff_words = populate_base_s2v_config( # <<<<<<<<<<<<<< * &w2v, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_6, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 441, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_6, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 440, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -3272,7 +3272,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 441, __pyx_L1_error) + __PYX_ERR(0, 440, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -3285,15 +3285,15 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_6); __Pyx_INCREF(__pyx_t_7); #else - __pyx_t_6 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 441, __pyx_L1_error) + __pyx_t_6 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 440, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); - __pyx_t_7 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 441, __pyx_L1_error) + __pyx_t_7 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 440, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_8 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_8)) __PYX_ERR(0, 441, __pyx_L1_error) + __pyx_t_8 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_8)) __PYX_ERR(0, 440, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_8); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_9 = Py_TYPE(__pyx_t_8)->tp_iternext; @@ -3301,7 +3301,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_6); index = 1; __pyx_t_7 = __pyx_t_9(__pyx_t_8); if (unlikely(!__pyx_t_7)) goto __pyx_L4_unpacking_failed; __Pyx_GOTREF(__pyx_t_7); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_9(__pyx_t_8), 2) < 0) __PYX_ERR(0, 441, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_9(__pyx_t_8), 2) < 0) __PYX_ERR(0, 440, __pyx_L1_error) __pyx_t_9 = NULL; __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; goto __pyx_L5_unpacking_done; @@ -3309,31 +3309,31 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; __pyx_t_9 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 441, __pyx_L1_error) + __PYX_ERR(0, 440, __pyx_L1_error) __pyx_L5_unpacking_done:; } - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 441, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 440, __pyx_L1_error) __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; - __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_7); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 441, __pyx_L1_error) + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_7); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 440, __pyx_L1_error) __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; __pyx_v_eff_sentences = __pyx_t_2; __pyx_v_eff_words = __pyx_t_10; - /* "pooling_inner.pyx":447 + /* "pooling_inner.pyx":446 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: * compute_base_sentence_pooling( */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 447, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 446, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 447, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 446, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_4 = ((!__pyx_t_5) != 0); if (__pyx_t_4) { - /* "pooling_inner.pyx":448 + /* "pooling_inner.pyx":447 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -3348,7 +3348,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":449 + /* "pooling_inner.pyx":448 * if not model.hierarchical: * with nogil: * compute_base_sentence_pooling( # <<<<<<<<<<<<<< @@ -3358,7 +3358,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_base_sentence_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":448 + /* "pooling_inner.pyx":447 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< @@ -3377,7 +3377,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } } - /* "pooling_inner.pyx":447 + /* "pooling_inner.pyx":446 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< @@ -3387,7 +3387,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L6; } - /* "pooling_inner.pyx":454 + /* "pooling_inner.pyx":453 * ) * else: * with nogil: # <<<<<<<<<<<<<< @@ -3403,7 +3403,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":455 + /* "pooling_inner.pyx":454 * else: * with nogil: * compute_base_sentence_hier_pooling( # <<<<<<<<<<<<<< @@ -3413,7 +3413,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences, __pyx_v_window_size, __pyx_v_window_stride); } - /* "pooling_inner.pyx":454 + /* "pooling_inner.pyx":453 * ) * else: * with nogil: # <<<<<<<<<<<<<< @@ -3434,7 +3434,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } __pyx_L6:; - /* "pooling_inner.pyx":438 + /* "pooling_inner.pyx":437 * cdef FTSentenceVecsConfig ft * * if not model.is_ft: # <<<<<<<<<<<<<< @@ -3444,7 +3444,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L3; } - /* "pooling_inner.pyx":462 + /* "pooling_inner.pyx":461 * ) * else: * init_ft_s2v_config(&ft, model, target, memory) # <<<<<<<<<<<<<< @@ -3452,31 +3452,31 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec * eff_sentences, eff_words = populate_ft_s2v_config( */ /*else*/ { - __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 462, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 461, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":466 + /* "pooling_inner.pyx":465 * eff_sentences, eff_words = populate_ft_s2v_config( * &ft, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 466, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 465, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_7 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 466, __pyx_L1_error) + __pyx_t_7 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 465, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":464 + /* "pooling_inner.pyx":463 * init_ft_s2v_config(&ft, model, target, memory) * * eff_sentences, eff_words = populate_ft_s2v_config( # <<<<<<<<<<<<<< * &ft, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_7, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 464, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_7, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 463, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -3485,7 +3485,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 464, __pyx_L1_error) + __PYX_ERR(0, 463, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -3498,15 +3498,15 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_7); __Pyx_INCREF(__pyx_t_6); #else - __pyx_t_7 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 464, __pyx_L1_error) + __pyx_t_7 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 463, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); - __pyx_t_6 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 464, __pyx_L1_error) + __pyx_t_6 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 463, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_8 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_8)) __PYX_ERR(0, 464, __pyx_L1_error) + __pyx_t_8 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_8)) __PYX_ERR(0, 463, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_8); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_9 = Py_TYPE(__pyx_t_8)->tp_iternext; @@ -3514,7 +3514,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_7); index = 1; __pyx_t_6 = __pyx_t_9(__pyx_t_8); if (unlikely(!__pyx_t_6)) goto __pyx_L13_unpacking_failed; __Pyx_GOTREF(__pyx_t_6); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_9(__pyx_t_8), 2) < 0) __PYX_ERR(0, 464, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_9(__pyx_t_8), 2) < 0) __PYX_ERR(0, 463, __pyx_L1_error) __pyx_t_9 = NULL; __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; goto __pyx_L14_unpacking_done; @@ -3522,36 +3522,36 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; __pyx_t_9 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 464, __pyx_L1_error) + __PYX_ERR(0, 463, __pyx_L1_error) __pyx_L14_unpacking_done:; } - __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_7); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 464, __pyx_L1_error) + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_7); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 463, __pyx_L1_error) __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 464, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 463, __pyx_L1_error) __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_v_eff_sentences = __pyx_t_10; __pyx_v_eff_words = __pyx_t_2; - /* "pooling_inner.pyx":470 + /* "pooling_inner.pyx":469 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: - * compute_ft_sentence_pooling(&ft, eff_sentences) + * compute_ft_sentence_pooling( */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 470, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 469, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 470, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 469, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_5 = ((!__pyx_t_4) != 0); if (__pyx_t_5) { - /* "pooling_inner.pyx":471 + /* "pooling_inner.pyx":470 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< - * compute_ft_sentence_pooling(&ft, eff_sentences) - * else: + * compute_ft_sentence_pooling( + * &ft, */ { #ifdef WITH_THREAD @@ -3561,22 +3561,22 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":472 + /* "pooling_inner.pyx":471 * if not model.hierarchical: * with nogil: - * compute_ft_sentence_pooling(&ft, eff_sentences) # <<<<<<<<<<<<<< - * else: - * with nogil: + * compute_ft_sentence_pooling( # <<<<<<<<<<<<<< + * &ft, + * eff_sentences */ __pyx_f_13pooling_inner_compute_ft_sentence_pooling((&__pyx_v_ft), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":471 + /* "pooling_inner.pyx":470 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< - * compute_ft_sentence_pooling(&ft, eff_sentences) - * else: + * compute_ft_sentence_pooling( + * &ft, */ /*finally:*/ { /*normal exit:*/{ @@ -3590,18 +3590,18 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } } - /* "pooling_inner.pyx":470 + /* "pooling_inner.pyx":469 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: - * compute_ft_sentence_pooling(&ft, eff_sentences) + * compute_ft_sentence_pooling( */ goto __pyx_L15; } - /* "pooling_inner.pyx":474 - * compute_ft_sentence_pooling(&ft, eff_sentences) + /* "pooling_inner.pyx":476 + * ) * else: * with nogil: # <<<<<<<<<<<<<< * compute_ft_sentence_hier_pooling( @@ -3616,7 +3616,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":475 + /* "pooling_inner.pyx":477 * else: * with nogil: * compute_ft_sentence_hier_pooling( # <<<<<<<<<<<<<< @@ -3626,8 +3626,8 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling((&__pyx_v_ft), __pyx_v_eff_sentences, __pyx_v_window_size, __pyx_v_window_stride); } - /* "pooling_inner.pyx":474 - * compute_ft_sentence_pooling(&ft, eff_sentences) + /* "pooling_inner.pyx":476 + * ) * else: * with nogil: # <<<<<<<<<<<<<< * compute_ft_sentence_hier_pooling( @@ -3649,7 +3649,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } __pyx_L3:; - /* "pooling_inner.pyx":482 + /* "pooling_inner.pyx":484 * ) * * return eff_sentences, eff_words # <<<<<<<<<<<<<< @@ -3657,11 +3657,11 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec * def init(): */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 482, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 484, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_6 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 482, __pyx_L1_error) + __pyx_t_6 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 484, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); - __pyx_t_7 = PyTuple_New(2); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 482, __pyx_L1_error) + __pyx_t_7 = PyTuple_New(2); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 484, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_7, 0, __pyx_t_1); @@ -3673,7 +3673,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_7 = 0; goto __pyx_L0; - /* "pooling_inner.pyx":402 + /* "pooling_inner.pyx":401 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< @@ -3695,7 +3695,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec return __pyx_r; } -/* "pooling_inner.pyx":484 +/* "pooling_inner.pyx":486 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -3723,7 +3723,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("init", 0); - /* "pooling_inner.pyx":485 + /* "pooling_inner.pyx":487 * * def init(): * return 1 # <<<<<<<<<<<<<< @@ -3735,7 +3735,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __pyx_r = __pyx_int_1; goto __pyx_L0; - /* "pooling_inner.pyx":484 + /* "pooling_inner.pyx":486 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -6350,26 +6350,26 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { __Pyx_GOTREF(__pyx_tuple__7); __Pyx_GIVEREF(__pyx_tuple__7); - /* "pooling_inner.pyx":402 + /* "pooling_inner.pyx":401 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_tuple__9 = PyTuple_Pack(10, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_window_size, __pyx_n_s_window_stride, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 402, __pyx_L1_error) + __pyx_tuple__9 = PyTuple_Pack(10, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_window_size, __pyx_n_s_window_stride, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 401, __pyx_L1_error) __Pyx_GOTREF(__pyx_tuple__9); __Pyx_GIVEREF(__pyx_tuple__9); - __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 10, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 402, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 402, __pyx_L1_error) + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 10, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 401, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 401, __pyx_L1_error) - /* "pooling_inner.pyx":484 + /* "pooling_inner.pyx":486 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 484, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 484, __pyx_L1_error) + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 486, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 486, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -6732,58 +6732,58 @@ if (!__Pyx_RefNanny) { if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_2) < 0) __PYX_ERR(0, 22, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":402 + /* "pooling_inner.pyx":401 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 402, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 401, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 402, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 401, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":484 + /* "pooling_inner.pyx":486 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 484, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 486, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 484, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 486, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":487 + /* "pooling_inner.pyx":489 * return 1 * * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 487, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 489, __pyx_L1_error) - /* "pooling_inner.pyx":488 + /* "pooling_inner.pyx":490 * * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 488, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 490, __pyx_L1_error) - /* "pooling_inner.pyx":489 + /* "pooling_inner.pyx":491 * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() # <<<<<<<<<<<<<< */ - __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 489, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 491, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 489, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 491, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 489, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 491, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; /* "pooling_inner.pyx":1 diff --git a/fse/test/test_base_s2v.py b/fse/test/test_base_s2v.py index 126ab8a..cc50142 100644 --- a/fse/test/test_base_s2v.py +++ b/fse/test/test_base_s2v.py @@ -436,7 +436,7 @@ def test_memory_creation(self): out = se._get_thread_working_mem() self.assertEqual(3, len(out)) self.assertEqual(se.sv.vector_size, len(out[0])) - self.assertEqual(se.sv.vector_size, len(out[2])) + self.assertEqual(se.sv.vector_size, len(out[1])) def test_infer_method(self): se = BaseSentence2VecModel(W2V) From f4aa40bf945c504791f518d9186d1137024b1fec Mon Sep 17 00:00:00 2001 From: OliverB Date: Tue, 3 Mar 2020 19:08:07 +0100 Subject: [PATCH 41/63] Readme --- README.md | 2 ++ 1 file changed, 2 insertions(+) diff --git a/README.md b/README.md index 5035c99..50e1912 100644 --- a/README.md +++ b/README.md @@ -11,6 +11,8 @@ Fast Sentence Embeddings (fse) Fast Sentence Embeddings is a Python library that serves as an addition to Gensim. This library is intended to compute *sentence vectors* for large collections of sentences or documents. +If you want to support fse, take a quick [survey](https://forms.gle/8uSU323fWUVtVwcAA) to improve it :-) + Features ------------ From d84b22566d7e7ad8263f50f104208f723525b88f Mon Sep 17 00:00:00 2001 From: OliverB Date: Fri, 22 May 2020 11:44:44 +0200 Subject: [PATCH 42/63] Minor Fixes --- fse/models/base_s2v.py | 6 + notebooks/STS-Benchmarks.ipynb | 604 ++++++++++++++++++------------ notebooks/Speed Benchmark.ipynb | 158 ++++++++ notebooks/Speed Comparision.ipynb | 152 ++------ notebooks/Tutorial.ipynb | 451 ++++------------------ 5 files changed, 645 insertions(+), 726 deletions(-) create mode 100644 notebooks/Speed Benchmark.ipynb diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index d92a187..966761e 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -159,6 +159,12 @@ def __init__( [X] compatibility [X] memory-usage --> Annoy + + [ ] unittests: + [ ] Check that custom index is numerical only + [ ] Raise + [ ] Unittest + """ set_madvise_for_mmap() diff --git a/notebooks/STS-Benchmarks.ipynb b/notebooks/STS-Benchmarks.ipynb index 13bb3bc..30091fc 100644 --- a/notebooks/STS-Benchmarks.ipynb +++ b/notebooks/STS-Benchmarks.ipynb @@ -36,7 +36,7 @@ "\n", "from gensim.models.keyedvectors import KeyedVectors, FastTextKeyedVectors\n", "\n", - "from fse.models import Average, SIF, uSIF\n", + "from fse.models import Average, SIF, uSIF, MaxPooling\n", "from fse import CSplitIndexedList\n", "\n", "from re import sub\n", @@ -221,58 +221,45 @@ }, { "cell_type": "code", - "execution_count": 8, + "execution_count": 25, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:38:54,395 : MainThread : INFO : loading Word2VecKeyedVectors object from /Volumes/Ext_HDD/Models/Static/glove.840B.300d.model\n", - "2019-09-11 10:38:59,173 : MainThread : INFO : loading vectors from /Volumes/Ext_HDD/Models/Static/glove.840B.300d.model.vectors.npy with mmap=None\n", - "2019-09-11 10:39:05,118 : MainThread : INFO : setting ignored attribute vectors_norm to None\n", - "2019-09-11 10:39:05,130 : MainThread : INFO : loaded /Volumes/Ext_HDD/Models/Static/glove.840B.300d.model\n", - "2019-09-11 10:39:05,138 : MainThread : INFO : loading pre-existing wv from /Users/oliverborchers/Library/Mobile Documents/com~apple~CloudDocs/Diss/Medium/Fast_Sentence_Embeddings/notebooks/data/glove_wv.vectors\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "Before memmap 2635050112\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "2019-09-11 10:39:05,544 : MainThread : INFO : no frequency mode: using wordfreq for estimation of frequency for language: en\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "After memmap 136\n" + "2020-02-16 11:58:42,025 : MainThread : INFO : loading Word2VecKeyedVectors object from /Volumes/Ext_HDD/Models/Static/glove.840B.300d.model\n", + "2020-02-16 11:58:46,648 : MainThread : INFO : loading vectors from /Volumes/Ext_HDD/Models/Static/glove.840B.300d.model.vectors.npy with mmap=r\n", + "2020-02-16 11:58:46,653 : MainThread : INFO : setting ignored attribute vectors_norm to None\n", + "2020-02-16 11:58:46,654 : MainThread : INFO : loaded /Volumes/Ext_HDD/Models/Static/glove.840B.300d.model\n", + "2020-02-16 11:58:46,657 : MainThread : INFO : no frequency mode: using wordfreq for estimation of frequency for language: en\n", + "2020-02-16 11:58:48,470 : MainThread : INFO : make sure you are using a model with valid word-frequency information. Otherwise use lang_freq argument.\n", + "2020-02-16 11:58:48,479 : MainThread : INFO : make sure you are using a model with valid word-frequency information. Otherwise use lang_freq argument.\n", + "/Users/oliverborchers/Library/Mobile Documents/com~apple~CloudDocs/Diss/Medium/Fast_Sentence_Embeddings/fse/models/base_s2v.py:172: UserWarning: C extension not loaded, training/inferring will be slow. Install a C compiler and reinstall fse.\n", + " \"C extension not loaded, training/inferring will be slow. \"\n" ] } ], "source": [ - "glove = KeyedVectors.load(path_to_models+\"glove.840B.300d.model\")\n", + "glove = KeyedVectors.load(path_to_models+\"glove.840B.300d.model\", mmap=\"r\")\n", "\n", - "print(f\"Before memmap {sys.getsizeof(glove.vectors)}\")\n", + "#print(f\"Before memmap {sys.getsizeof(glove.vectors)}\")\n", "\n", - "models[f\"CBOW-Glove\"] = Average(glove, wv_mapfile_path=\"data/glove\", lang_freq=\"en\")\n", + "#models[f\"CBOW-Glove\"] = Average(glove, wv_mapfile_path=\"data/glove\", lang_freq=\"en\")\n", "\n", - "print(f\"After memmap {sys.getsizeof(glove.vectors)}\")\n", + "#print(f\"After memmap {sys.getsizeof(glove.vectors)}\")\n", "\n", + "models[f\"CBOW-Glove\"] = Average(glove, lang_freq=\"en\")\n", "models[f\"SIF-Glove\"] = SIF(glove, components=15)\n", - "models[f\"uSIF-Glove\"] = uSIF(glove,length=11)" + "models[f\"uSIF-Glove\"] = uSIF(glove,length=11)\n", + "\n", + "models[f\"Max-Glove\"] = MaxPooling(glove)\n", + "models[f\"hMax-Glove\"] = MaxPooling(glove, hierarchical=True)" ] }, { "cell_type": "code", - "execution_count": 9, + "execution_count": 26, "metadata": {}, "outputs": [ { @@ -281,7 +268,7 @@ "True" ] }, - "execution_count": 9, + "execution_count": 26, "metadata": {}, "output_type": "execute_result" } @@ -300,18 +287,20 @@ }, { "cell_type": "code", - "execution_count": 10, + "execution_count": 8, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:39:16,246 : MainThread : INFO : loading Word2VecKeyedVectors object from /Volumes/Ext_HDD/Models/Static/google_news.model\n", - "2019-09-11 10:39:22,918 : MainThread : INFO : loading vectors from /Volumes/Ext_HDD/Models/Static/google_news.model.vectors.npy with mmap=r\n", - "2019-09-11 10:39:22,929 : MainThread : INFO : setting ignored attribute vectors_norm to None\n", - "2019-09-11 10:39:22,929 : MainThread : INFO : loaded /Volumes/Ext_HDD/Models/Static/google_news.model\n", - "2019-09-11 10:39:22,932 : MainThread : INFO : no frequency mode: using wordfreq for estimation of frequency for language: en\n" + "2020-02-16 11:48:12,816 : MainThread : INFO : loading Word2VecKeyedVectors object from /Volumes/Ext_HDD/Models/Static/google_news.model\n", + "2020-02-16 11:48:18,748 : MainThread : INFO : loading vectors from /Volumes/Ext_HDD/Models/Static/google_news.model.vectors.npy with mmap=r\n", + "2020-02-16 11:48:18,756 : MainThread : INFO : setting ignored attribute vectors_norm to None\n", + "2020-02-16 11:48:18,757 : MainThread : INFO : loaded /Volumes/Ext_HDD/Models/Static/google_news.model\n", + "2020-02-16 11:48:18,760 : MainThread : INFO : no frequency mode: using wordfreq for estimation of frequency for language: en\n", + "2020-02-16 11:48:21,162 : MainThread : INFO : make sure you are using a model with valid word-frequency information. Otherwise use lang_freq argument.\n", + "2020-02-16 11:48:21,165 : MainThread : INFO : make sure you are using a model with valid word-frequency information. Otherwise use lang_freq argument.\n" ] } ], @@ -320,28 +309,35 @@ "\n", "models[f\"CBOW-W2V\"] = Average(w2v, lang_freq=\"en\")\n", "models[f\"SIF-W2V\"] = SIF(w2v, components=10)\n", - "models[f\"uSIF-W2V\"] = uSIF(w2v, length=11)" + "models[f\"uSIF-W2V\"] = uSIF(w2v, length=11)\n", + "\n", + "models[f\"Max-W2V\"] = MaxPooling(w2v)\n", + "models[f\"hMax-W2V\"] = MaxPooling(w2v, hierarchical=True)" ] }, { "cell_type": "code", - "execution_count": 11, + "execution_count": 13, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:39:25,448 : MainThread : INFO : loading FastTextKeyedVectors object from /Volumes/Ext_HDD/Models/Static/ft_crawl_300d_2m.model\n", - "2019-09-11 10:39:32,132 : MainThread : INFO : loading vectors from /Volumes/Ext_HDD/Models/Static/ft_crawl_300d_2m.model.vectors.npy with mmap=r\n", - "2019-09-11 10:39:32,142 : MainThread : INFO : loading vectors_vocab from /Volumes/Ext_HDD/Models/Static/ft_crawl_300d_2m.model.vectors_vocab.npy with mmap=r\n", - "2019-09-11 10:39:32,154 : MainThread : INFO : loading vectors_ngrams from /Volumes/Ext_HDD/Models/Static/ft_crawl_300d_2m.model.vectors_ngrams.npy with mmap=r\n", - "2019-09-11 10:39:32,161 : MainThread : INFO : setting ignored attribute vectors_norm to None\n", - "2019-09-11 10:39:32,162 : MainThread : INFO : setting ignored attribute vectors_vocab_norm to None\n", - "2019-09-11 10:39:32,163 : MainThread : INFO : setting ignored attribute vectors_ngrams_norm to None\n", - "2019-09-11 10:39:32,164 : MainThread : INFO : setting ignored attribute buckets_word to None\n", - "2019-09-11 10:39:32,165 : MainThread : INFO : loaded /Volumes/Ext_HDD/Models/Static/ft_crawl_300d_2m.model\n", - "2019-09-11 10:39:32,168 : MainThread : INFO : no frequency mode: using wordfreq for estimation of frequency for language: en\n" + "2020-02-16 11:49:19,927 : MainThread : INFO : loading FastTextKeyedVectors object from /Volumes/Ext_HDD/Models/Static/ft_crawl_300d_2m.model\n", + "2020-02-16 11:49:24,020 : MainThread : INFO : loading vectors from /Volumes/Ext_HDD/Models/Static/ft_crawl_300d_2m.model.vectors.npy with mmap=r\n", + "2020-02-16 11:49:24,030 : MainThread : INFO : loading vectors_vocab from /Volumes/Ext_HDD/Models/Static/ft_crawl_300d_2m.model.vectors_vocab.npy with mmap=r\n", + "2020-02-16 11:49:24,036 : MainThread : INFO : loading vectors_ngrams from /Volumes/Ext_HDD/Models/Static/ft_crawl_300d_2m.model.vectors_ngrams.npy with mmap=r\n", + "2020-02-16 11:49:24,041 : MainThread : INFO : setting ignored attribute vectors_norm to None\n", + "2020-02-16 11:49:24,042 : MainThread : INFO : setting ignored attribute vectors_vocab_norm to None\n", + "2020-02-16 11:49:24,043 : MainThread : INFO : setting ignored attribute vectors_ngrams_norm to None\n", + "2020-02-16 11:49:24,043 : MainThread : INFO : setting ignored attribute buckets_word to None\n", + "2020-02-16 11:49:24,044 : MainThread : INFO : loaded /Volumes/Ext_HDD/Models/Static/ft_crawl_300d_2m.model\n", + "2020-02-16 11:49:24,047 : MainThread : INFO : no frequency mode: using wordfreq for estimation of frequency for language: en\n", + "2020-02-16 11:49:25,734 : MainThread : INFO : make sure you are using a model with valid word-frequency information. Otherwise use lang_freq argument.\n", + "2020-02-16 11:49:25,739 : MainThread : INFO : make sure you are using a model with valid word-frequency information. Otherwise use lang_freq argument.\n", + "/Users/oliverborchers/Library/Mobile Documents/com~apple~CloudDocs/Diss/Medium/Fast_Sentence_Embeddings/fse/models/base_s2v.py:172: UserWarning: C extension not loaded, training/inferring will be slow. Install a C compiler and reinstall fse.\n", + " \"C extension not loaded, training/inferring will be slow. \"\n" ] } ], @@ -349,23 +345,30 @@ "ft = FastTextKeyedVectors.load(path_to_models+\"ft_crawl_300d_2m.model\", mmap=\"r\")\n", "models[f\"CBOW-FT\"] = Average(ft, lang_freq=\"en\")\n", "models[f\"SIF-FT\"] = SIF(ft, components=10)\n", - "models[f\"uSIF-FT\"] = uSIF(ft, length=11)" + "models[f\"uSIF-FT\"] = uSIF(ft, length=11)\n", + "\n", + "models[f\"Max-FT\"] = MaxPooling(ft)\n", + "models[f\"hMax-FT\"] = MaxPooling(ft, hierarchical=True)" ] }, { "cell_type": "code", - "execution_count": 12, + "execution_count": 14, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:39:34,534 : MainThread : INFO : loading Word2VecKeyedVectors object from /Volumes/Ext_HDD/Models/Static/paranmt.model\n", - "2019-09-11 10:39:34,800 : MainThread : INFO : loading vectors from /Volumes/Ext_HDD/Models/Static/paranmt.model.vectors.npy with mmap=r\n", - "2019-09-11 10:39:34,811 : MainThread : INFO : setting ignored attribute vectors_norm to None\n", - "2019-09-11 10:39:34,812 : MainThread : INFO : loaded /Volumes/Ext_HDD/Models/Static/paranmt.model\n", - "2019-09-11 10:39:34,814 : MainThread : INFO : no frequency mode: using wordfreq for estimation of frequency for language: en\n" + "2020-02-16 11:49:42,202 : MainThread : INFO : loading Word2VecKeyedVectors object from /Volumes/Ext_HDD/Models/Static/paranmt.model\n", + "2020-02-16 11:49:42,390 : MainThread : INFO : loading vectors from /Volumes/Ext_HDD/Models/Static/paranmt.model.vectors.npy with mmap=r\n", + "2020-02-16 11:49:42,396 : MainThread : INFO : setting ignored attribute vectors_norm to None\n", + "2020-02-16 11:49:42,396 : MainThread : INFO : loaded /Volumes/Ext_HDD/Models/Static/paranmt.model\n", + "2020-02-16 11:49:42,399 : MainThread : INFO : no frequency mode: using wordfreq for estimation of frequency for language: en\n", + "2020-02-16 11:49:42,477 : MainThread : INFO : make sure you are using a model with valid word-frequency information. Otherwise use lang_freq argument.\n", + "2020-02-16 11:49:42,478 : MainThread : INFO : make sure you are using a model with valid word-frequency information. Otherwise use lang_freq argument.\n", + "/Users/oliverborchers/Library/Mobile Documents/com~apple~CloudDocs/Diss/Medium/Fast_Sentence_Embeddings/fse/models/base_s2v.py:172: UserWarning: C extension not loaded, training/inferring will be slow. Install a C compiler and reinstall fse.\n", + " \"C extension not loaded, training/inferring will be slow. \"\n" ] } ], @@ -374,23 +377,30 @@ "\n", "models[f\"CBOW-Paranmt\"] = Average(paranmt, lang_freq=\"en\")\n", "models[f\"SIF-Paranmt\"] = SIF(paranmt, components=10)\n", - "models[f\"uSIF-Paranmt\"] = uSIF(paranmt, length=11)" + "models[f\"uSIF-Paranmt\"] = uSIF(paranmt, length=11)\n", + "\n", + "models[f\"Max-Paranmt\"] = MaxPooling(paranmt)\n", + "models[f\"hMax-Paranmt\"] = MaxPooling(paranmt, hierarchical=True)" ] }, { "cell_type": "code", - "execution_count": 13, + "execution_count": 15, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:39:34,941 : MainThread : INFO : loading Word2VecKeyedVectors object from /Volumes/Ext_HDD/Models/Static/paragram_sl999_czeng.model\n", - "2019-09-11 10:39:35,099 : MainThread : INFO : loading vectors from /Volumes/Ext_HDD/Models/Static/paragram_sl999_czeng.model.vectors.npy with mmap=r\n", - "2019-09-11 10:39:35,108 : MainThread : INFO : setting ignored attribute vectors_norm to None\n", - "2019-09-11 10:39:35,109 : MainThread : INFO : loaded /Volumes/Ext_HDD/Models/Static/paragram_sl999_czeng.model\n", - "2019-09-11 10:39:35,111 : MainThread : INFO : no frequency mode: using wordfreq for estimation of frequency for language: en\n" + "2020-02-16 11:50:11,887 : MainThread : INFO : loading Word2VecKeyedVectors object from /Volumes/Ext_HDD/Models/Static/paragram_sl999_czeng.model\n", + "2020-02-16 11:50:11,990 : MainThread : INFO : loading vectors from /Volumes/Ext_HDD/Models/Static/paragram_sl999_czeng.model.vectors.npy with mmap=r\n", + "2020-02-16 11:50:11,994 : MainThread : INFO : setting ignored attribute vectors_norm to None\n", + "2020-02-16 11:50:11,995 : MainThread : INFO : loaded /Volumes/Ext_HDD/Models/Static/paragram_sl999_czeng.model\n", + "2020-02-16 11:50:11,998 : MainThread : INFO : no frequency mode: using wordfreq for estimation of frequency for language: en\n", + "2020-02-16 11:50:12,069 : MainThread : INFO : make sure you are using a model with valid word-frequency information. Otherwise use lang_freq argument.\n", + "2020-02-16 11:50:12,070 : MainThread : INFO : make sure you are using a model with valid word-frequency information. Otherwise use lang_freq argument.\n", + "/Users/oliverborchers/Library/Mobile Documents/com~apple~CloudDocs/Diss/Medium/Fast_Sentence_Embeddings/fse/models/base_s2v.py:172: UserWarning: C extension not loaded, training/inferring will be slow. Install a C compiler and reinstall fse.\n", + " \"C extension not loaded, training/inferring will be slow. \"\n" ] } ], @@ -399,7 +409,10 @@ "\n", "models[f\"CBOW-Paragram\"] = Average(paragram, lang_freq=\"en\")\n", "models[f\"SIF-Paragram\"] = SIF(paragram, components=10)\n", - "models[f\"uSIF-Paragram\"] = uSIF(paragram, length=11)" + "models[f\"uSIF-Paragram\"] = uSIF(paragram, length=11)\n", + "\n", + "models[f\"Max-Paragram\"] = MaxPooling(paragram)\n", + "models[f\"hMax-Paragram\"] = MaxPooling(paragram, hierarchical=True)" ] }, { @@ -418,7 +431,7 @@ }, { "cell_type": "code", - "execution_count": 14, + "execution_count": 16, "metadata": {}, "outputs": [], "source": [ @@ -433,222 +446,238 @@ }, { "cell_type": "code", - "execution_count": 15, + "execution_count": 17, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:39:35,222 : MainThread : INFO : scanning all indexed sentences and their word counts\n", - "2019-09-11 10:39:35,727 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:39:36,788 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 2195875 vocabulary: 2524 MB (2 GB)\n", - "2019-09-11 10:39:36,789 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:39:36,846 : MainThread : INFO : begin training\n", - "2019-09-11 10:39:37,353 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:39:37,354 : MainThread : INFO : training on 2758 effective sentences with 27351 effective words took 0s with 5430 sentences/s\n", - "2019-09-11 10:39:37,385 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:50:21,575 : MainThread : INFO : scanning all indexed sentences and their word counts\n", + "2020-02-16 11:50:21,945 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:50:22,962 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 3000000 vocabulary: 3447 MB (3 GB)\n", + "2020-02-16 11:50:22,963 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:50:23,020 : MainThread : INFO : begin training\n", + "2020-02-16 11:50:23,651 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:50:23,652 : MainThread : INFO : training on 2758 effective sentences with 23116 effective words took 0s with 4367 sentences/s\n", + "2020-02-16 11:50:23,683 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ - "CBOW-Glove 40.41\n" + "CBOW-W2V 61.54\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:39:37,798 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:39:38,648 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 2195875 vocabulary: 2524 MB (2 GB)\n", - "2019-09-11 10:39:38,649 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:39:38,662 : MainThread : INFO : pre-computing SIF weights for 2195875 words\n", - "2019-09-11 10:39:40,394 : MainThread : INFO : begin training\n", - "2019-09-11 10:39:40,830 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:39:40,887 : MainThread : INFO : computing 15 principal components took 0s\n", - "2019-09-11 10:39:40,889 : MainThread : INFO : removing 15 principal components took 0s\n", - "2019-09-11 10:39:40,890 : MainThread : INFO : training on 2758 effective sentences with 27351 effective words took 0s with 6316 sentences/s\n", - "2019-09-11 10:39:40,926 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:50:24,051 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:50:25,067 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 3000000 vocabulary: 3447 MB (3 GB)\n", + "2020-02-16 11:50:25,067 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:50:25,081 : MainThread : INFO : pre-computing SIF weights for 3000000 words\n", + "2020-02-16 11:50:27,027 : MainThread : INFO : begin training\n", + "2020-02-16 11:50:27,381 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:50:27,420 : MainThread : INFO : computing 10 principal components took 0s\n", + "2020-02-16 11:50:27,422 : MainThread : INFO : removing 10 principal components took 0s\n", + "2020-02-16 11:50:27,423 : MainThread : INFO : training on 2758 effective sentences with 23116 effective words took 0s with 7763 sentences/s\n", + "2020-02-16 11:50:27,460 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ - "SIF-Glove 71.95\n" + "SIF-W2V 71.12\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:39:41,411 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:39:42,249 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 2195875 vocabulary: 2524 MB (2 GB)\n", - "2019-09-11 10:39:42,249 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:39:42,264 : MainThread : INFO : pre-computing uSIF weights for 2195875 words\n", - "2019-09-11 10:39:50,589 : MainThread : INFO : begin training\n", - "2019-09-11 10:39:51,119 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:39:51,138 : MainThread : INFO : computing 5 principal components took 0s\n", - "2019-09-11 10:39:51,141 : MainThread : INFO : removing 5 principal components took 0s\n", - "2019-09-11 10:39:51,142 : MainThread : INFO : training on 2758 effective sentences with 27351 effective words took 0s with 5197 sentences/s\n", - "2019-09-11 10:39:51,186 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:50:27,821 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:50:28,819 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 3000000 vocabulary: 3447 MB (3 GB)\n", + "2020-02-16 11:50:28,820 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:50:28,832 : MainThread : INFO : pre-computing uSIF weights for 3000000 words\n", + "2020-02-16 11:50:37,657 : MainThread : INFO : begin training\n", + "2020-02-16 11:50:38,012 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:50:38,036 : MainThread : INFO : computing 5 principal components took 0s\n", + "2020-02-16 11:50:38,038 : MainThread : INFO : removing 5 principal components took 0s\n", + "2020-02-16 11:50:38,039 : MainThread : INFO : training on 2758 effective sentences with 23116 effective words took 0s with 7741 sentences/s\n", + "2020-02-16 11:50:38,078 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ - "uSIF-Glove 67.16\n" + "uSIF-W2V 66.99\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:39:51,643 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:39:53,870 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 3000000 vocabulary: 3447 MB (3 GB)\n", - "2019-09-11 10:39:53,871 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:39:53,952 : MainThread : INFO : begin training\n", - "2019-09-11 10:39:54,566 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:39:54,567 : MainThread : INFO : training on 2758 effective sentences with 23116 effective words took 0s with 4482 sentences/s\n", - "2019-09-11 10:39:54,606 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:50:38,522 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:50:39,531 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 3000000 vocabulary: 3447 MB (3 GB)\n", + "2020-02-16 11:50:39,532 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:50:39,587 : MainThread : INFO : begin training\n", + "2020-02-16 11:50:40,040 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:50:40,041 : MainThread : INFO : training on 2758 effective sentences with 23116 effective words took 0s with 6083 sentences/s\n", + "2020-02-16 11:50:40,066 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ - "CBOW-W2V 61.54\n" + "Max-W2V 66.52\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:39:55,064 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:39:56,280 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 3000000 vocabulary: 3447 MB (3 GB)\n", - "2019-09-11 10:39:56,280 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:39:56,294 : MainThread : INFO : pre-computing SIF weights for 3000000 words\n", - "2019-09-11 10:39:59,084 : MainThread : INFO : begin training\n", - "2019-09-11 10:39:59,549 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:39:59,570 : MainThread : INFO : computing 10 principal components took 0s\n", - "2019-09-11 10:39:59,573 : MainThread : INFO : removing 10 principal components took 0s\n", - "2019-09-11 10:39:59,574 : MainThread : INFO : training on 2758 effective sentences with 23116 effective words took 0s with 5922 sentences/s\n", - "2019-09-11 10:39:59,617 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:50:40,443 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:50:41,423 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 3000000 vocabulary: 3447 MB (3 GB)\n", + "2020-02-16 11:50:41,424 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:50:41,482 : MainThread : INFO : begin training\n", + "2020-02-16 11:50:42,563 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:50:42,564 : MainThread : INFO : training on 2758 effective sentences with 23116 effective words took 1s with 2549 sentences/s\n", + "2020-02-16 11:50:42,588 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ - "SIF-W2V 71.12\n" + "hMax-W2V 52.19\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:40:00,087 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:40:01,227 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 3000000 vocabulary: 3447 MB (3 GB)\n", - "2019-09-11 10:40:01,228 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:40:01,246 : MainThread : INFO : pre-computing uSIF weights for 3000000 words\n", - "2019-09-11 10:40:12,911 : MainThread : INFO : begin training\n", - "2019-09-11 10:40:13,382 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:40:13,403 : MainThread : INFO : computing 5 principal components took 0s\n", - "2019-09-11 10:40:13,407 : MainThread : INFO : removing 5 principal components took 0s\n", - "2019-09-11 10:40:13,408 : MainThread : INFO : training on 2758 effective sentences with 23116 effective words took 0s with 5839 sentences/s\n", - "2019-09-11 10:40:13,445 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:50:42,961 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:50:43,740 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 2000000 vocabulary: 6877 MB (6 GB)\n", + "2020-02-16 11:50:43,740 : MainThread : WARNING : The embeddings will likely not fit into RAM. Consider to use mapfile_path\n", + "2020-02-16 11:50:43,741 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:50:43,786 : MainThread : INFO : begin training\n", + "2020-02-16 11:50:44,844 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:50:44,845 : MainThread : INFO : training on 2758 effective sentences with 27528 effective words took 1s with 2604 sentences/s\n", + "2020-02-16 11:50:44,875 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ - "uSIF-W2V 66.99\n" + "CBOW-FT 48.49\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:40:13,890 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:40:15,745 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 2000000 vocabulary: 6877 MB (6 GB)\n", - "2019-09-11 10:40:15,746 : MainThread : WARNING : The embeddings will likely not fit into RAM. Consider to use mapfile_path\n", - "2019-09-11 10:40:15,747 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:40:15,804 : MainThread : INFO : begin training\n", - "2019-09-11 10:40:16,861 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:40:16,862 : MainThread : INFO : training on 2758 effective sentences with 27528 effective words took 1s with 2605 sentences/s\n", - "2019-09-11 10:40:16,894 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:50:45,288 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:50:46,090 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 2000000 vocabulary: 6877 MB (6 GB)\n", + "2020-02-16 11:50:46,090 : MainThread : WARNING : The embeddings will likely not fit into RAM. Consider to use mapfile_path\n", + "2020-02-16 11:50:46,091 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:50:46,106 : MainThread : INFO : pre-computing SIF weights for 2000000 words\n", + "2020-02-16 11:50:47,492 : MainThread : INFO : begin training\n", + "2020-02-16 11:50:47,868 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:50:47,904 : MainThread : INFO : computing 10 principal components took 0s\n", + "2020-02-16 11:50:47,906 : MainThread : INFO : removing 10 principal components took 0s\n", + "2020-02-16 11:50:47,907 : MainThread : INFO : training on 2758 effective sentences with 27528 effective words took 0s with 7314 sentences/s\n", + "2020-02-16 11:50:47,944 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ - "CBOW-FT 48.49\n" + "SIF-FT 73.38\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:40:17,317 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:40:18,202 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 2000000 vocabulary: 6877 MB (6 GB)\n", - "2019-09-11 10:40:18,203 : MainThread : WARNING : The embeddings will likely not fit into RAM. Consider to use mapfile_path\n", - "2019-09-11 10:40:18,204 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:40:18,221 : MainThread : INFO : pre-computing SIF weights for 2000000 words\n", - "2019-09-11 10:40:20,197 : MainThread : INFO : begin training\n", - "2019-09-11 10:40:20,713 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:40:20,735 : MainThread : INFO : computing 10 principal components took 0s\n", - "2019-09-11 10:40:20,737 : MainThread : INFO : removing 10 principal components took 0s\n", - "2019-09-11 10:40:20,738 : MainThread : INFO : training on 2758 effective sentences with 27528 effective words took 0s with 5331 sentences/s\n", - "2019-09-11 10:40:20,779 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:50:48,337 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:50:49,104 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 2000000 vocabulary: 6877 MB (6 GB)\n", + "2020-02-16 11:50:49,105 : MainThread : WARNING : The embeddings will likely not fit into RAM. Consider to use mapfile_path\n", + "2020-02-16 11:50:49,105 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:50:49,121 : MainThread : INFO : pre-computing uSIF weights for 2000000 words\n", + "2020-02-16 11:50:55,026 : MainThread : INFO : begin training\n", + "2020-02-16 11:50:55,385 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:50:55,415 : MainThread : INFO : computing 5 principal components took 0s\n", + "2020-02-16 11:50:55,418 : MainThread : INFO : removing 5 principal components took 0s\n", + "2020-02-16 11:50:55,419 : MainThread : INFO : training on 2758 effective sentences with 27528 effective words took 0s with 7664 sentences/s\n", + "2020-02-16 11:50:55,464 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ - "SIF-FT 73.38\n" + "uSIF-FT 69.40\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:40:21,219 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:40:22,061 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 2000000 vocabulary: 6877 MB (6 GB)\n", - "2019-09-11 10:40:22,062 : MainThread : WARNING : The embeddings will likely not fit into RAM. Consider to use mapfile_path\n", - "2019-09-11 10:40:22,063 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:40:22,078 : MainThread : INFO : pre-computing uSIF weights for 2000000 words\n", - "2019-09-11 10:40:30,034 : MainThread : INFO : begin training\n", - "2019-09-11 10:40:30,553 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:40:30,578 : MainThread : INFO : computing 5 principal components took 0s\n", - "2019-09-11 10:40:30,581 : MainThread : INFO : removing 5 principal components took 0s\n", - "2019-09-11 10:40:30,582 : MainThread : INFO : training on 2758 effective sentences with 27528 effective words took 0s with 5309 sentences/s\n", - "2019-09-11 10:40:30,624 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:50:55,835 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:50:56,624 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 2000000 vocabulary: 6877 MB (6 GB)\n", + "2020-02-16 11:50:56,624 : MainThread : WARNING : The embeddings will likely not fit into RAM. Consider to use mapfile_path\n", + "2020-02-16 11:50:56,625 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:50:56,669 : MainThread : INFO : begin training\n", + "2020-02-16 11:50:57,421 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:50:57,422 : MainThread : INFO : training on 2758 effective sentences with 27528 effective words took 0s with 3664 sentences/s\n", + "2020-02-16 11:50:57,446 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { "name": "stdout", "output_type": "stream", "text": [ - "uSIF-FT 69.40\n" + "Max-FT 57.10\n" + ] + }, + { + "name": "stderr", + "output_type": "stream", + "text": [ + "2020-02-16 11:50:57,802 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:50:58,587 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 2000000 vocabulary: 6877 MB (6 GB)\n", + "2020-02-16 11:50:58,588 : MainThread : WARNING : The embeddings will likely not fit into RAM. Consider to use mapfile_path\n", + "2020-02-16 11:50:58,589 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:50:58,634 : MainThread : INFO : begin training\n", + "2020-02-16 11:51:00,569 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:51:00,570 : MainThread : INFO : training on 2758 effective sentences with 27528 effective words took 1s with 1424 sentences/s\n", + "2020-02-16 11:51:00,593 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "hMax-FT 52.97\n" ] }, { "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:40:31,120 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:40:31,190 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", - "2019-09-11 10:40:31,191 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:40:31,206 : MainThread : INFO : begin training\n", - "2019-09-11 10:40:31,889 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:40:31,890 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 4030 sentences/s\n", - "2019-09-11 10:40:31,921 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:51:00,947 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:51:00,975 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", + "2020-02-16 11:51:00,976 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:51:00,992 : MainThread : INFO : begin training\n", + "2020-02-16 11:51:01,621 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:51:01,621 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 4379 sentences/s\n", + "2020-02-16 11:51:01,646 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { @@ -662,16 +691,16 @@ "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:40:32,297 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:40:32,327 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", - "2019-09-11 10:40:32,328 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:40:32,340 : MainThread : INFO : pre-computing SIF weights for 77224 words\n", - "2019-09-11 10:40:32,396 : MainThread : INFO : begin training\n", - "2019-09-11 10:40:32,772 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:40:32,792 : MainThread : INFO : computing 10 principal components took 0s\n", - "2019-09-11 10:40:32,794 : MainThread : INFO : removing 10 principal components took 0s\n", - "2019-09-11 10:40:32,795 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 7319 sentences/s\n", - "2019-09-11 10:40:32,832 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:51:02,004 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:51:02,029 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", + "2020-02-16 11:51:02,030 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:51:02,043 : MainThread : INFO : pre-computing SIF weights for 77224 words\n", + "2020-02-16 11:51:02,090 : MainThread : INFO : begin training\n", + "2020-02-16 11:51:02,456 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:51:02,487 : MainThread : INFO : computing 10 principal components took 0s\n", + "2020-02-16 11:51:02,489 : MainThread : INFO : removing 10 principal components took 0s\n", + "2020-02-16 11:51:02,490 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 7522 sentences/s\n", + "2020-02-16 11:51:02,525 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { @@ -685,16 +714,16 @@ "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:40:33,253 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:40:33,282 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", - "2019-09-11 10:40:33,283 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:40:33,296 : MainThread : INFO : pre-computing uSIF weights for 77224 words\n", - "2019-09-11 10:40:33,533 : MainThread : INFO : begin training\n", - "2019-09-11 10:40:33,964 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:40:33,992 : MainThread : INFO : computing 5 principal components took 0s\n", - "2019-09-11 10:40:33,994 : MainThread : INFO : removing 5 principal components took 0s\n", - "2019-09-11 10:40:33,995 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 6374 sentences/s\n", - "2019-09-11 10:40:34,043 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:51:02,892 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:51:02,919 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", + "2020-02-16 11:51:02,920 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:51:02,933 : MainThread : INFO : pre-computing uSIF weights for 77224 words\n", + "2020-02-16 11:51:03,162 : MainThread : INFO : begin training\n", + "2020-02-16 11:51:03,526 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:51:03,552 : MainThread : INFO : computing 5 principal components took 0s\n", + "2020-02-16 11:51:03,555 : MainThread : INFO : removing 5 principal components took 0s\n", + "2020-02-16 11:51:03,556 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 7553 sentences/s\n", + "2020-02-16 11:51:03,594 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { @@ -708,13 +737,53 @@ "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:40:34,572 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:40:34,650 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", - "2019-09-11 10:40:34,651 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:40:34,662 : MainThread : INFO : begin training\n", - "2019-09-11 10:40:35,184 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:40:35,185 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 5270 sentences/s\n", - "2019-09-11 10:40:35,214 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:51:03,959 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:51:03,986 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", + "2020-02-16 11:51:03,987 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:51:04,000 : MainThread : INFO : begin training\n", + "2020-02-16 11:51:04,479 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:51:04,480 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 5747 sentences/s\n", + "2020-02-16 11:51:04,506 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Max-Paranmt 71.57\n" + ] + }, + { + "name": "stderr", + "output_type": "stream", + "text": [ + "2020-02-16 11:51:04,870 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:51:04,902 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", + "2020-02-16 11:51:04,902 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:51:04,921 : MainThread : INFO : begin training\n", + "2020-02-16 11:51:06,129 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:51:06,130 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 1s with 2280 sentences/s\n", + "2020-02-16 11:51:06,154 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "hMax-Paranmt 54.62\n" + ] + }, + { + "name": "stderr", + "output_type": "stream", + "text": [ + "2020-02-16 11:51:06,520 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:51:06,548 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", + "2020-02-16 11:51:06,548 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:51:06,564 : MainThread : INFO : begin training\n", + "2020-02-16 11:51:07,058 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:51:07,059 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 5575 sentences/s\n", + "2020-02-16 11:51:07,083 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { @@ -728,16 +797,16 @@ "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:40:35,613 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:40:35,648 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", - "2019-09-11 10:40:35,649 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:40:35,665 : MainThread : INFO : pre-computing SIF weights for 77224 words\n", - "2019-09-11 10:40:35,724 : MainThread : INFO : begin training\n", - "2019-09-11 10:40:36,294 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:40:36,319 : MainThread : INFO : computing 10 principal components took 0s\n", - "2019-09-11 10:40:36,322 : MainThread : INFO : removing 10 principal components took 0s\n", - "2019-09-11 10:40:36,323 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 4828 sentences/s\n", - "2019-09-11 10:40:36,364 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + "2020-02-16 11:51:07,449 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:51:07,475 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", + "2020-02-16 11:51:07,476 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:51:07,489 : MainThread : INFO : pre-computing SIF weights for 77224 words\n", + "2020-02-16 11:51:07,539 : MainThread : INFO : begin training\n", + "2020-02-16 11:51:07,912 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:51:07,942 : MainThread : INFO : computing 10 principal components took 0s\n", + "2020-02-16 11:51:07,945 : MainThread : INFO : removing 10 principal components took 0s\n", + "2020-02-16 11:51:07,946 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 7367 sentences/s\n", + "2020-02-16 11:51:07,982 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { @@ -751,15 +820,16 @@ "name": "stderr", "output_type": "stream", "text": [ - "2019-09-11 10:40:36,831 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", - "2019-09-11 10:40:36,859 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", - "2019-09-11 10:40:36,860 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", - "2019-09-11 10:40:36,871 : MainThread : INFO : pre-computing uSIF weights for 77224 words\n", - "2019-09-11 10:40:37,130 : MainThread : INFO : begin training\n", - "2019-09-11 10:40:37,598 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2019-09-11 10:40:37,626 : MainThread : INFO : computing 5 principal components took 0s\n", - "2019-09-11 10:40:37,628 : MainThread : INFO : removing 5 principal components took 0s\n", - "2019-09-11 10:40:37,629 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 5878 sentences/s\n" + "2020-02-16 11:51:08,355 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:51:08,381 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", + "2020-02-16 11:51:08,382 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:51:08,393 : MainThread : INFO : pre-computing uSIF weights for 77224 words\n", + "2020-02-16 11:51:08,621 : MainThread : INFO : begin training\n", + "2020-02-16 11:51:08,979 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:51:09,006 : MainThread : INFO : computing 5 principal components took 0s\n", + "2020-02-16 11:51:09,008 : MainThread : INFO : removing 5 principal components took 0s\n", + "2020-02-16 11:51:09,009 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 7692 sentences/s\n", + "2020-02-16 11:51:09,049 : MainThread : INFO : scanning all indexed sentences and their word counts\n" ] }, { @@ -768,6 +838,45 @@ "text": [ "uSIF-Paragram 73.64\n" ] + }, + { + "name": "stderr", + "output_type": "stream", + "text": [ + "2020-02-16 11:51:09,429 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:51:09,458 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", + "2020-02-16 11:51:09,459 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:51:09,475 : MainThread : INFO : begin training\n", + "2020-02-16 11:51:09,952 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:51:09,953 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 0s with 5772 sentences/s\n", + "2020-02-16 11:51:09,978 : MainThread : INFO : scanning all indexed sentences and their word counts\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "Max-Paragram 59.82\n" + ] + }, + { + "name": "stderr", + "output_type": "stream", + "text": [ + "2020-02-16 11:51:10,344 : MainThread : INFO : finished scanning 2758 sentences with an average length of 9 and 27528 total words\n", + "2020-02-16 11:51:10,371 : MainThread : INFO : estimated memory for 2758 sentences with 300 dimensions and 77224 vocabulary: 91 MB (0 GB)\n", + "2020-02-16 11:51:10,372 : MainThread : INFO : initializing sentence vectors for 2758 sentences\n", + "2020-02-16 11:51:10,388 : MainThread : INFO : begin training\n", + "2020-02-16 11:51:11,659 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-16 11:51:11,660 : MainThread : INFO : training on 2758 effective sentences with 27441 effective words took 1s with 2167 sentences/s\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "hMax-Paragram 50.55\n" + ] } ], "source": [ @@ -782,7 +891,7 @@ }, { "cell_type": "code", - "execution_count": 16, + "execution_count": 18, "metadata": {}, "outputs": [ { @@ -811,18 +920,6 @@ " \n", " \n", " \n", - " CBOW-Glove\n", - " 40.41\n", - " \n", - " \n", - " SIF-Glove\n", - " 71.95\n", - " \n", - " \n", - " uSIF-Glove\n", - " 67.16\n", - " \n", - " \n", " CBOW-W2V\n", " 61.54\n", " \n", @@ -835,6 +932,14 @@ " 66.99\n", " \n", " \n", + " Max-W2V\n", + " 66.52\n", + " \n", + " \n", + " hMax-W2V\n", + " 52.19\n", + " \n", + " \n", " CBOW-FT\n", " 48.49\n", " \n", @@ -847,6 +952,14 @@ " 69.40\n", " \n", " \n", + " Max-FT\n", + " 57.10\n", + " \n", + " \n", + " hMax-FT\n", + " 52.97\n", + " \n", + " \n", " CBOW-Paranmt\n", " 79.85\n", " \n", @@ -859,6 +972,14 @@ " 79.02\n", " \n", " \n", + " Max-Paranmt\n", + " 71.57\n", + " \n", + " \n", + " hMax-Paranmt\n", + " 54.62\n", + " \n", + " \n", " CBOW-Paragram\n", " 50.38\n", " \n", @@ -870,30 +991,43 @@ " uSIF-Paragram\n", " 73.64\n", " \n", + " \n", + " Max-Paragram\n", + " 59.82\n", + " \n", + " \n", + " hMax-Paragram\n", + " 50.55\n", + " \n", " \n", "\n", "" ], "text/plain": [ " Pearson\n", - "CBOW-Glove 40.41\n", - "SIF-Glove 71.95\n", - "uSIF-Glove 67.16\n", "CBOW-W2V 61.54\n", "SIF-W2V 71.12\n", "uSIF-W2V 66.99\n", + "Max-W2V 66.52\n", + "hMax-W2V 52.19\n", "CBOW-FT 48.49\n", "SIF-FT 73.38\n", "uSIF-FT 69.40\n", + "Max-FT 57.10\n", + "hMax-FT 52.97\n", "CBOW-Paranmt 79.85\n", "SIF-Paranmt 76.75\n", "uSIF-Paranmt 79.02\n", + "Max-Paranmt 71.57\n", + "hMax-Paranmt 54.62\n", "CBOW-Paragram 50.38\n", "SIF-Paragram 73.86\n", - "uSIF-Paragram 73.64" + "uSIF-Paragram 73.64\n", + "Max-Paragram 59.82\n", + "hMax-Paragram 50.55" ] }, - "execution_count": 16, + "execution_count": 18, "metadata": {}, "output_type": "execute_result" } diff --git a/notebooks/Speed Benchmark.ipynb b/notebooks/Speed Benchmark.ipynb new file mode 100644 index 0000000..31e98f7 --- /dev/null +++ b/notebooks/Speed Benchmark.ipynb @@ -0,0 +1,158 @@ +{ + "cells": [ + { + "cell_type": "code", + "execution_count": 1, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "2020-02-27 13:36:31,584 : MainThread : INFO : loading projection weights from /Users/oliverborchers/gensim-data/glove-wiki-gigaword-100/glove-wiki-gigaword-100.gz\n", + "2020-02-27 13:37:17,383 : MainThread : INFO : loaded (400000, 100) matrix from /Users/oliverborchers/gensim-data/glove-wiki-gigaword-100/glove-wiki-gigaword-100.gz\n", + "/Users/oliverborchers/anaconda3/envs/fsedev/lib/python3.7/site-packages/smart_open/smart_open_lib.py:402: UserWarning: This function is deprecated, use smart_open.open instead. See the migration notes for details: https://github.com/RaRe-Technologies/smart_open/blob/master/README.rst#migrating-to-the-new-open-function\n", + " 'See the migration notes for details: %s' % _MIGRATION_NOTES_URL\n" + ] + }, + { + "name": "stdout", + "output_type": "stream", + "text": [ + "6468640\n" + ] + } + ], + "source": [ + "from fse.models import Average\n", + "from fse.models.average import train_average_np\n", + "from fse.models.average_inner import train_average_cy\n", + "from fse.models.pooling import train_pooling_np\n", + "\n", + "from fse.models.average import MAX_WORDS_IN_BATCH\n", + "\n", + "from fse import IndexedList\n", + "\n", + "import numpy as np\n", + "\n", + "import logging\n", + "logging.basicConfig(format='%(asctime)s : %(threadName)s : %(levelname)s : %(message)s', level=logging.INFO)\n", + "\n", + "import gensim.downloader as api\n", + "data = api.load(\"quora-duplicate-questions\")\n", + "glove = api.load(\"glove-wiki-gigaword-100\")\n", + "\n", + "def process_sentence(sentence):\n", + " \"\"\"\n", + " Here it is important to insert a whitespace before ?.\n", + " Otherwise tokens like \"bitcoin?\" will be treated as one word, which is out of vocabulary.\n", + " \"\"\"\n", + " \n", + " sentence = sentence.lower()\n", + " sentence = sentence.replace(\"?\", \" ?\")\n", + " sentence = sentence.replace(\"!\", \" !\")\n", + " sentence = sentence.replace(\".\", \" .\")\n", + " return sentence.split()\n", + "\n", + "sentences = []\n", + "for d in data:\n", + " # Let's blow up the data a bit by replicating each sentence.\n", + " for i in range(8):\n", + " sentences.append(process_sentence(d[\"question1\"]))\n", + " sentences.append(process_sentence(d[\"question2\"]))\n", + "s = IndexedList(sentences)\n", + "print(len(s))" + ] + }, + { + "cell_type": "code", + "execution_count": 3, + "metadata": {}, + "outputs": [ + { + "name": "stdout", + "output_type": "stream", + "text": [ + "10000\n", + "1\n" + ] + } + ], + "source": [ + "from fse.models.pooling import FAST_VERSION, MAX_WORDS_IN_BATCH\n", + "print(MAX_WORDS_IN_BATCH)\n", + "print(FAST_VERSION)" + ] + }, + { + "cell_type": "code", + "execution_count": 8, + "metadata": {}, + "outputs": [ + { + "name": "stderr", + "output_type": "stream", + "text": [ + "2020-02-27 13:42:30,093 : MainThread : INFO : scanning all indexed sentences and their word counts\n", + "2020-02-27 13:42:35,094 : MainThread : INFO : SCANNING : finished 3356210 sentences with 40926093 words\n", + "2020-02-27 13:42:39,354 : MainThread : WARNING : found 16 empty sentences\n", + "2020-02-27 13:42:39,355 : MainThread : INFO : finished scanning 6468640 sentences with an average length of 12 and 78941152 total words\n", + "2020-02-27 13:42:39,538 : MainThread : INFO : estimated memory for 6468640 sentences with 100 dimensions and 400000 vocabulary: 2621 MB (2 GB)\n", + "2020-02-27 13:42:39,539 : MainThread : INFO : initializing sentence vectors for 6468640 sentences\n", + "2020-02-27 13:43:03,685 : MainThread : INFO : begin training\n", + "2020-02-27 13:43:08,705 : MainThread : INFO : PROGRESS : finished 21.53% with 1392680 sentences and 16341256 words, 278536 sentences/s\n", + "2020-02-27 13:43:13,709 : MainThread : INFO : PROGRESS : finished 43.75% with 2830125 sentences and 33228900 words, 287489 sentences/s\n", + "2020-02-27 13:43:18,712 : MainThread : INFO : PROGRESS : finished 71.59% with 4631171 sentences and 54389632 words, 360209 sentences/s\n", + "2020-02-27 13:43:23,715 : MainThread : INFO : PROGRESS : finished 97.88% with 6331766 sentences and 74366291 words, 340119 sentences/s\n", + "2020-02-27 13:43:24,274 : MainThread : INFO : worker thread finished; awaiting finish of 1 more threads\n", + "2020-02-27 13:43:24,280 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", + "2020-02-27 13:43:24,284 : MainThread : INFO : training on 6468624 effective sentences with 75978256 effective words took 20s with 314033 sentences/s\n" + ] + }, + { + "data": { + "text/plain": [ + "(6468624, 75978256)" + ] + }, + "execution_count": 8, + "metadata": {}, + "output_type": "execute_result" + } + ], + "source": [ + "from fse.models import MaxPooling\n", + "model = MaxPooling(glove, workers=2)\n", + "model.train(s)" + ] + }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [] + } + ], + "metadata": { + "kernelspec": { + "display_name": "Python 3", + "language": "python", + "name": "python3" + }, + "language_info": { + "codemirror_mode": { + "name": "ipython", + "version": 3 + }, + "file_extension": ".py", + "mimetype": "text/x-python", + "name": "python", + "nbconvert_exporter": "python", + "pygments_lexer": "ipython3", + "version": "3.7.3" + } + }, + "nbformat": 4, + "nbformat_minor": 2 +} diff --git a/notebooks/Speed Comparision.ipynb b/notebooks/Speed Comparision.ipynb index 8df2ece..0fcf741 100644 --- a/notebooks/Speed Comparision.ipynb +++ b/notebooks/Speed Comparision.ipynb @@ -9,22 +9,14 @@ }, { "cell_type": "code", - "execution_count": 1, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "/Users/oliverborchers/anaconda3/envs/fsedev/lib/python3.7/site-packages/smart_open/smart_open_lib.py:398: UserWarning: This function is deprecated, use smart_open.open instead. See the migration notes for details: https://github.com/RaRe-Technologies/smart_open/blob/master/README.rst#migrating-to-the-new-open-function\n", - " 'See the migration notes for details: %s' % _MIGRATION_NOTES_URL\n" - ] - } - ], + "outputs": [], "source": [ "from fse.models import Average\n", "from fse.models.average import train_average_np\n", "from fse.models.average_inner import train_average_cy\n", + "from fse.models.pooling import train_pooling_np\n", "\n", "from fse.models.average import MAX_WORDS_IN_BATCH\n", "\n", @@ -54,14 +46,14 @@ }, { "cell_type": "code", - "execution_count": 2, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ "import gensim.downloader as api\n", "\n", "w2v = api.load(\"glove-wiki-gigaword-100\")\n", - "ft = api.load(\"fasttext-wiki-news-subwords-300\")" + "#ft = api.load(\"fasttext-wiki-news-subwords-300\")" ] }, { @@ -77,20 +69,9 @@ }, { "cell_type": "code", - "execution_count": 3, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "1" - ] - }, - "execution_count": 3, - "metadata": {}, - "output_type": "execute_result" - } - ], + "outputs": [], "source": [ "from fse.models.average import FAST_VERSION\n", "FAST_VERSION" @@ -98,17 +79,9 @@ }, { "cell_type": "code", - "execution_count": 4, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "300 µs ± 18.8 µs per loop (mean ± std. dev. of 7 runs, 1000 loops each)\n" - ] - } - ], + "outputs": [], "source": [ "%%timeit\n", "w2v_avg = Average(w2v)" @@ -116,17 +89,9 @@ }, { "cell_type": "code", - "execution_count": 5, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "363 ms ± 9.95 ms per loop (mean ± std. dev. of 7 runs, 1 loop each)\n" - ] - } - ], + "outputs": [], "source": [ "%%timeit\n", "w2v_avg = Average(w2v, lang_freq=\"en\")" @@ -141,11 +106,13 @@ }, { "cell_type": "code", - "execution_count": 6, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ "w2v_avg = Average(w2v)\n", + "w2v_pool = MaxPooling(w2v)\n", + "\n", "statistics = w2v_avg.scan_sentences(sentences)\n", "w2v_avg.prep.prepare_vectors(sv=w2v_avg.sv, total_sentences=statistics[\"max_index\"], update=False)\n", "memory = w2v_avg._get_thread_working_mem()" @@ -153,17 +120,9 @@ }, { "cell_type": "code", - "execution_count": 7, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "34.3 ms ± 2.16 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n" - ] - } - ], + "outputs": [], "source": [ "%%timeit\n", "train_average_np(model=w2v_avg, indexed_sentences=sentences, target=w2v_avg.sv.vectors, memory=memory)" @@ -171,22 +130,25 @@ }, { "cell_type": "code", - "execution_count": 8, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "2.94 ms ± 120 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n" - ] - } - ], + "outputs": [], "source": [ "%%timeit\n", "train_average_cy(model=w2v_avg, indexed_sentences=sentences, target=w2v_avg.sv.vectors, memory=memory)" ] }, + { + "cell_type": "code", + "execution_count": null, + "metadata": {}, + "outputs": [], + "source": [ + "%%timeit\n", + "\n", + "train_pooling_np(model=w2v_avg, indexed_sentences=sentences, target=w2v_avg.sv.vectors, memory=memory)" + ] + }, { "cell_type": "markdown", "metadata": {}, @@ -196,20 +158,9 @@ }, { "cell_type": "code", - "execution_count": 9, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "True" - ] - }, - "execution_count": 9, - "metadata": {}, - "output_type": "execute_result" - } - ], + "outputs": [], "source": [ "out_w2v_np = np.zeros_like(w2v_avg.sv.vectors)\n", "out_w2v_cy = np.zeros_like(w2v_avg.sv.vectors)\n", @@ -228,7 +179,7 @@ }, { "cell_type": "code", - "execution_count": 10, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ @@ -240,17 +191,9 @@ }, { "cell_type": "code", - "execution_count": 11, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "42.1 ms ± 4.21 ms per loop (mean ± std. dev. of 7 runs, 10 loops each)\n" - ] - } - ], + "outputs": [], "source": [ "%%timeit\n", "train_average_np(model=ft_avg, indexed_sentences=sentences, target=ft_avg.sv.vectors, memory=memory)" @@ -258,17 +201,9 @@ }, { "cell_type": "code", - "execution_count": 12, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "4.14 ms ± 207 µs per loop (mean ± std. dev. of 7 runs, 100 loops each)\n" - ] - } - ], + "outputs": [], "source": [ "%%timeit\n", "train_average_cy(model=ft_avg, indexed_sentences=sentences, target=ft_avg.sv.vectors, memory=memory)" @@ -283,20 +218,9 @@ }, { "cell_type": "code", - "execution_count": 13, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "True" - ] - }, - "execution_count": 13, - "metadata": {}, - "output_type": "execute_result" - } - ], + "outputs": [], "source": [ "out_ft_np = np.zeros_like(ft_avg.sv.vectors)\n", "out_ft_cy = np.zeros_like(ft_avg.sv.vectors)\n", diff --git a/notebooks/Tutorial.ipynb b/notebooks/Tutorial.ipynb index 11a1c88..35838ae 100644 --- a/notebooks/Tutorial.ipynb +++ b/notebooks/Tutorial.ipynb @@ -310,8 +310,8 @@ "name": "stderr", "output_type": "stream", "text": [ - "2020-02-10 21:42:23,353 : MainThread : INFO : loading projection weights from /Users/oliverborchers/gensim-data/glove-wiki-gigaword-100/glove-wiki-gigaword-100.gz\n", - "2020-02-10 21:43:05,757 : MainThread : INFO : loaded (400000, 100) matrix from /Users/oliverborchers/gensim-data/glove-wiki-gigaword-100/glove-wiki-gigaword-100.gz\n" + "2020-02-17 16:56:46,057 : MainThread : INFO : loading projection weights from /Users/oliverborchers/gensim-data/glove-wiki-gigaword-100/glove-wiki-gigaword-100.gz\n", + "2020-02-17 16:57:30,853 : MainThread : INFO : loaded (400000, 100) matrix from /Users/oliverborchers/gensim-data/glove-wiki-gigaword-100/glove-wiki-gigaword-100.gz\n" ] } ], @@ -343,12 +343,24 @@ } ], "source": [ + "def process_sentence(sentence):\n", + " \"\"\"\n", + " Here it is important to insert a whitespace before ?.\n", + " Otherwise tokens like \"bitcoin?\" will be treated as one word, which is out of vocabulary.\n", + " \"\"\"\n", + " \n", + " sentence = sentence.lower()\n", + " sentence = sentence.replace(\"?\", \" ?\")\n", + " sentence = sentence.replace(\"!\", \" !\")\n", + " sentence = sentence.replace(\".\", \" .\")\n", + " return sentence.split()\n", + "\n", "sentences = []\n", "for d in data:\n", " # Let's blow up the data a bit by replicating each sentence.\n", " for i in range(8):\n", - " sentences.append(d[\"question1\"].split())\n", - " sentences.append(d[\"question2\"].split())\n", + " sentences.append(process_sentence(d[\"question1\"]))\n", + " sentences.append(process_sentence(d[\"question2\"]))\n", "s = IndexedList(sentences)\n", "print(len(s))" ] @@ -383,58 +395,53 @@ }, { "cell_type": "code", - "execution_count": 13, + "execution_count": 15, "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "2020-02-10 21:43:26,025 : MainThread : INFO : no frequency mode: using wordfreq for estimation of frequency for language: en\n" - ] - } - ], + "outputs": [], "source": [ - "from fse.models import uSIF\n", - "model = uSIF(glove, workers=2, lang_freq=\"en\")" + "from fse.models import SIF\n", + "model = SIF(glove, workers=1, lang_freq=\"en\")" ] }, { "cell_type": "code", - "execution_count": 14, + "execution_count": 16, "metadata": {}, "outputs": [ { "name": "stderr", "output_type": "stream", "text": [ - "2020-02-10 21:43:26,444 : MainThread : INFO : scanning all indexed sentences and their word counts\n", - "2020-02-10 21:43:31,445 : MainThread : INFO : SCANNING : finished 4413745 sentences with 48816338 words\n", - "2020-02-10 21:43:33,734 : MainThread : WARNING : found 16 empty sentences\n", - "2020-02-10 21:43:33,735 : MainThread : INFO : finished scanning 6468640 sentences with an average length of 11 and 71556728 total words\n", - "2020-02-10 21:43:33,851 : MainThread : INFO : estimated memory for 6468640 sentences with 100 dimensions and 400000 vocabulary: 2621 MB (2 GB)\n", - "2020-02-10 21:43:33,852 : MainThread : INFO : initializing sentence vectors for 6468640 sentences\n", - "2020-02-10 21:43:56,067 : MainThread : INFO : pre-computing uSIF weights for 400000 words\n", - "2020-02-10 21:43:57,262 : MainThread : INFO : begin training\n", - "2020-02-10 21:44:02,273 : MainThread : INFO : PROGRESS : finished 37.80% with 2444873 sentences and 18587179 words, 488974 sentences/s\n", - "2020-02-10 21:44:07,274 : MainThread : INFO : PROGRESS : finished 75.57% with 4888132 sentences and 37203558 words, 488651 sentences/s\n", - "2020-02-10 21:44:10,545 : MainThread : INFO : worker thread finished; awaiting finish of 1 more threads\n", - "2020-02-10 21:44:10,546 : MainThread : INFO : worker thread finished; awaiting finish of 0 more threads\n", - "2020-02-10 21:44:10,547 : MainThread : INFO : sampling 2677165 vectors to compute principal components\n", - "2020-02-10 21:44:18,377 : MainThread : INFO : computing 5 principal components took 7s\n", - "2020-02-10 21:44:26,036 : MainThread : INFO : removing 5 principal components took 7s\n", - "2020-02-10 21:44:26,037 : MainThread : INFO : training on 6468624 effective sentences with 49255184 effective words took 13s with 486929 sentences/s\n" + "2020-02-17 16:59:44,054 : MainThread : INFO : scanning all indexed sentences and their word counts\n", + "2020-02-17 16:59:49,056 : MainThread : INFO : SCANNING : finished 4242578 sentences with 51760509 words\n", + "2020-02-17 16:59:51,664 : MainThread : WARNING : found 16 empty sentences\n", + "2020-02-17 16:59:51,664 : MainThread : INFO : finished scanning 6468640 sentences with an average length of 12 and 78941152 total words\n", + "2020-02-17 16:59:51,817 : MainThread : INFO : estimated memory for 6468640 sentences with 100 dimensions and 400000 vocabulary: 2621 MB (2 GB)\n", + "2020-02-17 16:59:51,818 : MainThread : INFO : initializing sentence vectors for 6468640 sentences\n", + "2020-02-17 17:00:16,694 : MainThread : INFO : begin training\n", + "2020-02-17 17:00:22,290 : MainThread : INFO : PROGRESS : finished 1.90% with 122760 sentences and 1444012 words, 24552 sentences/s\n", + "2020-02-17 17:00:27,661 : MainThread : INFO : PROGRESS : finished 3.93% with 254215 sentences and 2983767 words, 26291 sentences/s\n", + "2020-02-17 17:00:33,315 : MainThread : INFO : PROGRESS : finished 5.70% with 369018 sentences and 4331506 words, 22960 sentences/s\n", + "2020-02-17 17:00:38,388 : MainThread : INFO : PROGRESS : finished 7.23% with 467548 sentences and 5486200 words, 19706 sentences/s\n", + "2020-02-17 17:00:43,589 : MainThread : INFO : PROGRESS : finished 9.01% with 582797 sentences and 6833403 words, 23049 sentences/s\n", + "2020-02-17 17:00:48,756 : MainThread : INFO : PROGRESS : finished 10.78% with 697570 sentences and 8181704 words, 22954 sentences/s\n" ] }, { - "data": { - "text/plain": [ - "(6468624, 49255184)" - ] - }, - "execution_count": 14, - "metadata": {}, - "output_type": "execute_result" + "ename": "KeyboardInterrupt", + "evalue": "", + "output_type": "error", + "traceback": [ + "\u001b[0;31m---------------------------------------------------------------------------\u001b[0m", + "\u001b[0;31mKeyboardInterrupt\u001b[0m Traceback (most recent call last)", + "\u001b[0;32m\u001b[0m in \u001b[0;36m\u001b[0;34m\u001b[0m\n\u001b[0;32m----> 1\u001b[0;31m \u001b[0mmodel\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mtrain\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0ms\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m", + "\u001b[0;32m~/Library/Mobile Documents/com~apple~CloudDocs/Diss/Medium/Fast_Sentence_Embeddings/fse/models/base_s2v.py\u001b[0m in \u001b[0;36mtrain\u001b[0;34m(self, sentences, update, queue_factor, report_delay)\u001b[0m\n\u001b[1;32m 778\u001b[0m \u001b[0mtotal_sentences\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mstatistics\u001b[0m\u001b[0;34m[\u001b[0m\u001b[0;34m\"total_sentences\"\u001b[0m\u001b[0;34m]\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 779\u001b[0m \u001b[0mqueue_factor\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mqueue_factor\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 780\u001b[0;31m \u001b[0mreport_delay\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mreport_delay\u001b[0m\u001b[0;34m,\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 781\u001b[0m )\n\u001b[1;32m 782\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n", + "\u001b[0;32m~/Library/Mobile Documents/com~apple~CloudDocs/Diss/Medium/Fast_Sentence_Embeddings/fse/models/base_s2v.py\u001b[0m in \u001b[0;36m_train_manager\u001b[0;34m(self, data_iterable, total_sentences, queue_factor, report_delay)\u001b[0m\n\u001b[1;32m 878\u001b[0m \u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 879\u001b[0m jobs, eff_sentences, eff_words = self._log_train_progress(\n\u001b[0;32m--> 880\u001b[0;31m \u001b[0mprogress_queue\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mtotal_sentences\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mtotal_sentences\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0mreport_delay\u001b[0m\u001b[0;34m=\u001b[0m\u001b[0mreport_delay\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 881\u001b[0m )\n\u001b[1;32m 882\u001b[0m \u001b[0;32mreturn\u001b[0m \u001b[0mjobs\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0meff_sentences\u001b[0m\u001b[0;34m,\u001b[0m \u001b[0meff_words\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", + "\u001b[0;32m~/Library/Mobile Documents/com~apple~CloudDocs/Diss/Medium/Fast_Sentence_Embeddings/fse/models/base_s2v.py\u001b[0m in \u001b[0;36m_log_train_progress\u001b[0;34m(self, progress_queue, total_sentences, report_delay)\u001b[0m\n\u001b[1;32m 979\u001b[0m \u001b[0msentence_inc\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 980\u001b[0m \u001b[0;32mwhile\u001b[0m \u001b[0munfinished_worker_count\u001b[0m \u001b[0;34m>\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 981\u001b[0;31m \u001b[0mreport\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0mprogress_queue\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mget\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 982\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mreport\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0;31m# a thread reporting that it finished\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 983\u001b[0m \u001b[0munfinished_worker_count\u001b[0m \u001b[0;34m-=\u001b[0m \u001b[0;36m1\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", + "\u001b[0;32m~/anaconda3/envs/fsedev/lib/python3.7/queue.py\u001b[0m in \u001b[0;36mget\u001b[0;34m(self, block, timeout)\u001b[0m\n\u001b[1;32m 168\u001b[0m \u001b[0;32melif\u001b[0m \u001b[0mtimeout\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 169\u001b[0m \u001b[0;32mwhile\u001b[0m \u001b[0;32mnot\u001b[0m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0m_qsize\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 170\u001b[0;31m \u001b[0mself\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mnot_empty\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0mwait\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 171\u001b[0m \u001b[0;32melif\u001b[0m \u001b[0mtimeout\u001b[0m \u001b[0;34m<\u001b[0m \u001b[0;36m0\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 172\u001b[0m \u001b[0;32mraise\u001b[0m \u001b[0mValueError\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m\"'timeout' must be a non-negative number\"\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", + "\u001b[0;32m~/anaconda3/envs/fsedev/lib/python3.7/threading.py\u001b[0m in \u001b[0;36mwait\u001b[0;34m(self, timeout)\u001b[0m\n\u001b[1;32m 294\u001b[0m \u001b[0;32mtry\u001b[0m\u001b[0;34m:\u001b[0m \u001b[0;31m# restore state no matter what (e.g., KeyboardInterrupt)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 295\u001b[0m \u001b[0;32mif\u001b[0m \u001b[0mtimeout\u001b[0m \u001b[0;32mis\u001b[0m \u001b[0;32mNone\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0;32m--> 296\u001b[0;31m \u001b[0mwaiter\u001b[0m\u001b[0;34m.\u001b[0m\u001b[0macquire\u001b[0m\u001b[0;34m(\u001b[0m\u001b[0;34m)\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[0m\u001b[1;32m 297\u001b[0m \u001b[0mgotit\u001b[0m \u001b[0;34m=\u001b[0m \u001b[0;32mTrue\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n\u001b[1;32m 298\u001b[0m \u001b[0;32melse\u001b[0m\u001b[0;34m:\u001b[0m\u001b[0;34m\u001b[0m\u001b[0;34m\u001b[0m\u001b[0m\n", + "\u001b[0;31mKeyboardInterrupt\u001b[0m: " + ] } ], "source": [ @@ -457,62 +464,9 @@ }, { "cell_type": "code", - "execution_count": 15, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "2020-02-10 21:44:26,044 : MainThread : INFO : scanning all indexed sentences and their word counts\n", - "2020-02-10 21:44:26,045 : MainThread : INFO : finished scanning 1 sentences with an average length of 3 and 3 total words\n", - "2020-02-10 21:44:26,047 : MainThread : INFO : removing 5 principal components took 0s\n" - ] - }, - { - "data": { - "text/plain": [ - "array([[ 2.58718699e-01, -2.96060964e-02, 2.75402740e-02,\n", - " -2.84367323e-01, -7.66123906e-02, 4.69245732e-01,\n", - " -1.08420335e-01, 2.74900701e-02, -6.51940107e-02,\n", - " -3.48900437e-01, -3.30639817e-03, -7.38123357e-02,\n", - " 1.99272603e-01, 1.58340886e-01, -1.19580366e-01,\n", - " -2.94115573e-01, 9.44712311e-02, -1.60182863e-01,\n", - " -3.77932310e-01, 3.62254620e-01, -1.04730584e-01,\n", - " 2.72801578e-01, -3.65233980e-02, -1.77455202e-01,\n", - " 1.13285437e-01, 9.37283933e-02, -2.23851919e-01,\n", - " -5.82970530e-02, 4.76750970e-01, 1.19097173e-01,\n", - " 2.51136065e-01, 2.99976945e-01, 3.93524468e-01,\n", - " 1.26966879e-01, 1.19876862e-03, 2.52949506e-01,\n", - " 1.83217332e-01, 6.29579574e-02, 2.79819459e-01,\n", - " -1.32508770e-01, -1.32991910e-01, 1.35885537e-01,\n", - " 2.27139533e-01, -1.15716822e-01, -1.42301470e-01,\n", - " -1.17087245e-01, -4.09713805e-01, 3.27361971e-01,\n", - " 4.02728885e-01, -1.03995442e-01, -1.11777350e-01,\n", - " -2.22199619e-01, 8.88717845e-02, 1.40038133e-01,\n", - " 9.33110416e-02, -4.57023978e-01, 5.33070415e-04,\n", - " 2.35848241e-02, -1.01058990e-01, 1.92017779e-01,\n", - " 5.89273870e-02, 6.50007427e-01, 3.70745771e-02,\n", - " -1.76800102e-01, 5.68467081e-02, -8.34784359e-02,\n", - " 2.63056427e-01, 2.63333827e-01, 2.41475731e-01,\n", - " -1.44381002e-02, -1.83939964e-01, 1.32494122e-01,\n", - " 4.73914444e-02, -5.11802197e-01, -2.29695290e-01,\n", - " 4.10918087e-01, 3.23362201e-01, -2.24917591e-01,\n", - " -4.87098396e-02, 8.45644325e-02, -5.20223975e-02,\n", - " 4.39105839e-01, 3.65090221e-01, -2.31675640e-01,\n", - " -5.31933427e-01, -1.05184980e-01, -3.58922482e-01,\n", - " 1.48164511e-01, -2.76803404e-01, -2.75730073e-01,\n", - " -8.09157640e-02, 7.01733604e-02, 5.82995176e-01,\n", - " -2.55794972e-02, -4.16485399e-01, -1.49305463e-01,\n", - " -2.40142912e-01, 1.64687201e-01, -7.57704377e-02,\n", - " 2.89778203e-01]], dtype=float32)" - ] - }, - "execution_count": 15, - "metadata": {}, - "output_type": "execute_result" - } - ], + "outputs": [], "source": [ "tmp = (\"Hello my friends\".split(), 0)\n", "model.infer([tmp])" @@ -534,40 +488,9 @@ }, { "cell_type": "code", - "execution_count": 16, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "array([ 0.06564295, 0.0012124 , 0.02864488, 0.29741746, 0.16618916,\n", - " -0.33291832, -0.25267577, -0.11754846, -0.00645616, -0.0986203 ,\n", - " -0.03115754, 0.11605997, -0.06554113, -0.26570198, -0.19048208,\n", - " -0.05398345, -0.00800271, 0.06935053, 0.02384207, 0.15339501,\n", - " 0.0931268 , 0.04639681, -0.23096606, 0.1496515 , -0.14506361,\n", - " 0.02416093, 0.05317958, 0.06964332, -0.07533754, 0.006847 ,\n", - " -0.2449986 , 0.22674319, -0.09827837, -0.09429546, 0.13742915,\n", - " 0.15489256, 0.20663384, -0.10573711, -0.09373225, -0.21597916,\n", - " -0.04622186, -0.07917423, 0.03237222, -0.09423919, -0.24972957,\n", - " 0.1362891 , -0.24006578, 0.05784579, -0.06796119, -0.49378857,\n", - " -0.22550753, -0.00635221, -0.03531939, 0.2966177 , -0.17845063,\n", - " -0.5473429 , -0.14862986, -0.03552294, 0.6726266 , -0.07657065,\n", - " 0.05169982, -0.18650085, -0.1508371 , -0.00102763, 0.05002424,\n", - " 0.14072034, -0.19600302, 0.21199626, 0.12934232, -0.07024553,\n", - " 0.07888867, -0.06151189, -0.25036365, -0.30167574, -0.13264754,\n", - " -0.16540082, -0.18127926, 0.08294387, -0.10994744, -0.19283013,\n", - " 0.21519725, 0.23384331, -0.06928872, -0.08626016, 0.00445604,\n", - " -0.03812613, 0.05476346, -0.11798044, -0.03109498, -0.11745913,\n", - " -0.01188265, 0.03115901, 0.02107314, -0.32339057, -0.26852998,\n", - " 0.10675619, 0.2325929 , -0.2140762 , 0.15390754, 0.26329693],\n", - " dtype=float32)" - ] - }, - "execution_count": 16, - "metadata": {}, - "output_type": "execute_result" - } - ], + "outputs": [], "source": [ "model.sv[0]" ] @@ -581,18 +504,9 @@ }, { "cell_type": "code", - "execution_count": 17, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "0.965\n", - "0.035\n" - ] - } - ], + "outputs": [], "source": [ "print(model.sv.similarity(0,1).round(3))\n", "print(model.sv.distance(0,1).round(3))" @@ -607,53 +521,18 @@ }, { "cell_type": "code", - "execution_count": 18, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "(['Should', 'I', 'buy', 'tiago?'], 100)\n" - ] - } - ], + "outputs": [], "source": [ "print(s[100])" ] }, { "cell_type": "code", - "execution_count": 19, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "2020-02-10 21:44:26,077 : MainThread : INFO : precomputing L2-norms of sentence vectors\n" - ] - }, - { - "data": { - "text/plain": [ - "[(3727921, 1.0),\n", - " (1807119, 1.0),\n", - " (3727935, 1.0),\n", - " (3727933, 1.0),\n", - " (3727931, 1.0),\n", - " (3727929, 1.0),\n", - " (3727927, 1.0),\n", - " (3727925, 1.0),\n", - " (3727923, 1.0),\n", - " (599388, 1.0)]" - ] - }, - "execution_count": 19, - "metadata": {}, - "output_type": "execute_result" - } - ], + "outputs": [], "source": [ "model.sv.most_similar(100)\n", "# Division by zero can happen if you encounter empy sentences" @@ -668,33 +547,20 @@ }, { "cell_type": "code", - "execution_count": 20, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "[(['Should', 'I', 'buy', 'KTM', 'Duke', '200?'], 3727921, 1.0),\n", - " (['Should', 'I', 'buy', 'Xiaomi', 'Redmi', 'Note', '3?', 'Why?'],\n", - " 1807119,\n", - " 1.0),\n", - " (['Should', 'I', 'buy', 'KTM', 'Duke', '200?'], 3727935, 1.0),\n", - " (['Should', 'I', 'buy', 'KTM', 'Duke', '200?'], 3727933, 1.0),\n", - " (['Should', 'I', 'buy', 'KTM', 'Duke', '200?'], 3727931, 1.0),\n", - " (['Should', 'I', 'buy', 'KTM', 'Duke', '200?'], 3727929, 1.0),\n", - " (['Should', 'I', 'buy', 'KTM', 'Duke', '200?'], 3727927, 1.0),\n", - " (['Should', 'I', 'buy', 'KTM', 'Duke', '200?'], 3727925, 1.0),\n", - " (['Should', 'I', 'buy', 'KTM', 'Duke', '200?'], 3727923, 1.0),\n", - " (['Should', 'I', 'buy', 'bitcoin?'], 599388, 1.0)]" - ] - }, - "execution_count": 20, - "metadata": {}, - "output_type": "execute_result" - } - ], + "outputs": [], "source": [ - "model.sv.most_similar(100, indexable=s.items)" + "index = 100\n", + "print(\"Input: \")\n", + "print(\" \".join(s.items[index]))\n", + "print(\"Output: \")\n", + "\n", + "for data in model.sv.most_similar(index, indexable=s.items):\n", + " sentence = \" \".join(data[0])\n", + " print(\n", + " f\"Idx: {data[1]:7d} \\t Sim: {data[2]:1.3f} \\t Sentence: {sentence}\"\n", + " )" ] }, { @@ -713,129 +579,9 @@ }, { "cell_type": "code", - "execution_count": 21, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "data": { - "text/plain": [ - "[(['Which',\n", - " 'is',\n", - " 'more',\n", - " 'easy',\n", - " 'to',\n", - " 'learn?',\n", - " 'Ruby',\n", - " 'on',\n", - " 'Rails',\n", - " 'or',\n", - " 'Python/Django?'],\n", - " 4717071,\n", - " 0.9463648796081543),\n", - " (['Which',\n", - " 'is',\n", - " 'more',\n", - " 'easy',\n", - " 'to',\n", - " 'learn?',\n", - " 'Ruby',\n", - " 'on',\n", - " 'Rails',\n", - " 'or',\n", - " 'Python/Django?'],\n", - " 4717059,\n", - " 0.9463648796081543),\n", - " (['Which',\n", - " 'is',\n", - " 'more',\n", - " 'easy',\n", - " 'to',\n", - " 'learn?',\n", - " 'Ruby',\n", - " 'on',\n", - " 'Rails',\n", - " 'or',\n", - " 'Python/Django?'],\n", - " 4717063,\n", - " 0.9463648796081543),\n", - " (['Which',\n", - " 'is',\n", - " 'more',\n", - " 'easy',\n", - " 'to',\n", - " 'learn?',\n", - " 'Ruby',\n", - " 'on',\n", - " 'Rails',\n", - " 'or',\n", - " 'Python/Django?'],\n", - " 4717067,\n", - " 0.9463648796081543),\n", - " (['Which',\n", - " 'is',\n", - " 'more',\n", - " 'easy',\n", - " 'to',\n", - " 'learn?',\n", - " 'Ruby',\n", - " 'on',\n", - " 'Rails',\n", - " 'or',\n", - " 'Python/Django?'],\n", - " 4717061,\n", - " 0.9463648796081543),\n", - " (['Which',\n", - " 'is',\n", - " 'more',\n", - " 'easy',\n", - " 'to',\n", - " 'learn?',\n", - " 'Ruby',\n", - " 'on',\n", - " 'Rails',\n", - " 'or',\n", - " 'Python/Django?'],\n", - " 4717065,\n", - " 0.9463648796081543),\n", - " (['Which',\n", - " 'is',\n", - " 'more',\n", - " 'easy',\n", - " 'to',\n", - " 'learn?',\n", - " 'Ruby',\n", - " 'on',\n", - " 'Rails',\n", - " 'or',\n", - " 'Python/Django?'],\n", - " 4717057,\n", - " 0.9463648796081543),\n", - " (['Which',\n", - " 'is',\n", - " 'more',\n", - " 'easy',\n", - " 'to',\n", - " 'learn?',\n", - " 'Ruby',\n", - " 'on',\n", - " 'Rails',\n", - " 'or',\n", - " 'Python/Django?'],\n", - " 4717069,\n", - " 0.9463648796081543),\n", - " (['How', 'can', 'I', 'make', 'some', 'easy', 'money?'],\n", - " 4405380,\n", - " 0.9432926177978516),\n", - " (['How', 'can', 'I', 'make', 'some', 'easy', 'money?'],\n", - " 5443135,\n", - " 0.9432926177978516)]" - ] - }, - "execution_count": 21, - "metadata": {}, - "output_type": "execute_result" - } - ], + "outputs": [], "source": [ "model.sv.similar_by_word(\"easy\", wv=glove, indexable=s.items)" ] @@ -849,58 +595,9 @@ }, { "cell_type": "code", - "execution_count": 22, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "2020-02-10 21:44:35,595 : MainThread : INFO : scanning all indexed sentences and their word counts\n", - "2020-02-10 21:44:35,597 : MainThread : INFO : finished scanning 1 sentences with an average length of 6 and 6 total words\n", - "2020-02-10 21:44:35,600 : MainThread : INFO : removing 5 principal components took 0s\n" - ] - }, - { - "data": { - "text/plain": [ - "[(['How', 'do', 'I', 'learn', 'Python', 'in', 'easy', 'way?'],\n", - " 6255666,\n", - " 0.9860048294067383),\n", - " (['How', 'do', 'I', 'learn', 'Python', 'in', 'easy', 'way?'],\n", - " 6255668,\n", - " 0.9860048294067383),\n", - " (['How', 'do', 'I', 'learn', 'Python', 'in', 'easy', 'way?'],\n", - " 418236,\n", - " 0.9860048294067383),\n", - " (['How', 'do', 'I', 'learn', 'Python', 'in', 'easy', 'way?'],\n", - " 418238,\n", - " 0.9860048294067383),\n", - " (['How', 'do', 'I', 'learn', 'Python', 'in', 'easy', 'way?'],\n", - " 6255664,\n", - " 0.9860048294067383),\n", - " (['How', 'do', 'I', 'learn', 'Python', 'in', 'easy', 'way?'],\n", - " 418232,\n", - " 0.9860048294067383),\n", - " (['How', 'do', 'I', 'learn', 'Python', 'in', 'easy', 'way?'],\n", - " 6255674,\n", - " 0.9860048294067383),\n", - " (['How', 'do', 'I', 'learn', 'Python', 'in', 'easy', 'way?'],\n", - " 6255672,\n", - " 0.9860048294067383),\n", - " (['How', 'do', 'I', 'learn', 'Python', 'in', 'easy', 'way?'],\n", - " 418230,\n", - " 0.9860048294067383),\n", - " (['How', 'do', 'I', 'learn', 'Python', 'in', 'easy', 'way?'],\n", - " 6255670,\n", - " 0.9860048294067383)]" - ] - }, - "execution_count": 22, - "metadata": {}, - "output_type": "execute_result" - } - ], + "outputs": [], "source": [ "model.sv.similar_by_sentence(\"Is this really easy to learn\".split(), model=model, indexable=s.items)" ] From b3e766e8ae0694cef6b73f8e0347d04c0a4d01ed Mon Sep 17 00:00:00 2001 From: OliverB Date: Fri, 22 May 2020 11:45:24 +0200 Subject: [PATCH 43/63] todo --- fse/models/base_s2v.py | 1 + 1 file changed, 1 insertion(+) diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index 966761e..2ffadac 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -151,6 +151,7 @@ def __init__( [ ] z-score transformation is quite nice [ ] sentencevectors: + [ ] remove the L1 normalization or make it optional? [X] similar_by_sentence model type check [ ] approximate NN search for large files [X] compare ANN libraries From 4fac67028ebfbc57c3a87937ab1465ec288af3f4 Mon Sep 17 00:00:00 2001 From: OliverB Date: Fri, 22 May 2020 13:48:16 +0200 Subject: [PATCH 44/63] Refactored unittests --- fse/models/utils.py | 51 ++- fse/test/shared_imports.py | 43 +++ fse/test/test_average.py | 103 ++--- fse/test/test_base_s2v.py | 2 +- fse/test/test_pooling.py | 768 +++++++++++++++++++------------------ fse/test/test_utils.py | 24 +- 6 files changed, 537 insertions(+), 454 deletions(-) create mode 100644 fse/test/shared_imports.py diff --git a/fse/models/utils.py b/fse/models/utils.py index 32fc826..da0dc09 100644 --- a/fse/models/utils.py +++ b/fse/models/utils.py @@ -7,7 +7,9 @@ from sklearn.decomposition import TruncatedSVD -from numpy import ndarray, float32 as REAL, ones, vstack, inf as INF, dtype +from gensim.models.utils_any2vec import ft_ngram_hashes + +from numpy import ndarray, float32 as REAL, ones, vstack, inf as INF, dtype, amax as np_amax, zeros, sum as np_sum from numpy.random import choice from time import time @@ -20,6 +22,53 @@ logger = logging.getLogger(__name__) +def get_ft_word_vector( + word: str, + model, + ) -> ndarray: + """ Function to compute the FT vectors + + Parameters + ---------- + word : str + String representation of token + + Returns + ------- + ndarray + FT vector representation + + """ + size = model.wv.vector_size + vocab = model.wv.vocab + + w_vectors = model.wv.vectors + w_weights = model.word_weights + + max_ngrams = model.batch_ngrams + min_n = model.wv.min_n + max_n = model.wv.max_n + bucket = model.wv.bucket + + oov_weight = np_amax(w_weights) + + ngram_vectors = model.wv.vectors_ngrams + + if word in vocab: + vocab_index = vocab[word].index + return w_vectors[vocab_index] * w_weights[vocab_index] + else: + # Requires additional temporary storage + ngram_hashes = ft_ngram_hashes(word, min_n, max_n, bucket, True)[ + :max_ngrams + ] + if len(ngram_hashes) == 0: + return zeros(size, dtype=REAL) + return ( + oov_weight + * np_sum(ngram_vectors[ngram_hashes], axis=0) + / len(ngram_hashes) + ) def set_madvise_for_mmap(return_madvise: bool = False) -> object: """ Method used to set madvise parameters. diff --git a/fse/test/shared_imports.py b/fse/test/shared_imports.py new file mode 100644 index 0000000..3578c5a --- /dev/null +++ b/fse/test/shared_imports.py @@ -0,0 +1,43 @@ +import logging +import unittest + +import numpy as np + +from pathlib import Path + +from gensim.models import Word2Vec, FastText + +from fse.models.base_s2v import EPS + +logger = logging.getLogger(__name__) + +# Global objs +CORPUS = Path("fse/test/test_data/test_sentences.txt") +with open(CORPUS, "r") as f: + SENTENCES = [l.split() for i, l in enumerate(f)] + +# Models +DIM = 5 + +# Deterministic W2V +W2V_DET = Word2Vec(min_count=1, size=DIM) +W2V_DET.build_vocab(SENTENCES) +W2V_DET.wv.vectors[:,] = np.arange(len(W2V_DET.wv.vectors), dtype=np.float32)[:, None] + +# Random W2V +W2V_RNG = Word2Vec(min_count=1, size=DIM) +W2V_RNG.build_vocab(SENTENCES) + +# Deterministic FT +FT_DET = FastText(min_count=1, size=DIM) +FT_DET.build_vocab(SENTENCES) +FT_DET.wv.vectors = FT_DET.wv.vectors_vocab = np.ones_like( + FT_DET.wv.vectors, dtype=np.float32 +) +FT_DET.wv.vectors_ngrams[:,] = np.arange(len(FT_DET.wv.vectors_ngrams), dtype=np.float32)[ + :, None +] + +# Random FT +FT_RNG = FastText(min_count=1, size=DIM) +FT_RNG.build_vocab(SENTENCES) diff --git a/fse/test/test_average.py b/fse/test/test_average.py index 33f4090..d95c108 100644 --- a/fse/test/test_average.py +++ b/fse/test/test_average.py @@ -9,28 +9,9 @@ Automated tests for checking the average model. """ -import logging -import unittest - -from pathlib import Path - -import numpy as np +from shared_imports import * from fse.models.average import Average, train_average_np -from fse.models.base_s2v import EPS - -from gensim.models import Word2Vec, FastText - -logger = logging.getLogger(__name__) - -CORPUS = Path("fse/test/test_data/test_sentences.txt") -DIM = 5 -W2V = Word2Vec(min_count=1, size=DIM) -with open(CORPUS, "r") as f: - SENTENCES = [l.split() for i, l in enumerate(f)] -W2V.build_vocab(SENTENCES) -W2V.wv.vectors[:,] = np.arange(len(W2V.wv.vectors), dtype=np.float32)[:, None] - class TestAverageFunctions(unittest.TestCase): def setUp(self): @@ -41,7 +22,7 @@ def setUp(self): ["pull", "12345678910111213"], ] self.sentences = [(s, i) for i, s in enumerate(self.sentences)] - self.model = Average(W2V) + self.model = Average(W2V_DET) self.model.prep.prepare_vectors( sv=self.model.sv, total_sentences=len(self.sentences), update=False ) @@ -58,7 +39,7 @@ def test_cython(self): self.assertEqual(10000, MAX_WORDS_IN_BATCH) self.assertEqual(40, MAX_NGRAMS_IN_BATCH) - def test_average_train_np_w2v(self): + def test_average_train_np_w2v_det(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() output = train_average_np( @@ -69,7 +50,7 @@ def test_average_train_np_w2v(self): self.assertTrue((164.5 == self.model.sv[1]).all()) self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) - def test_average_train_cy_w2v(self): + def test_average_train_cy_w2v_det(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() @@ -83,36 +64,29 @@ def test_average_train_cy_w2v(self): self.assertTrue((164.5 == self.model.sv[1]).all()) self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) - def test_average_train_np_ft(self): - ft = FastText(min_count=1, size=DIM) - ft.build_vocab(SENTENCES) - m = Average(ft) + def test_average_train_np_ft_det(self): + m = Average(FT_DET) m.prep.prepare_vectors( sv=m.sv, total_sentences=len(self.sentences), update=False ) m._pre_train_calls() - m.wv.vectors = m.wv.vectors_vocab = np.ones_like(m.wv.vectors, dtype=np.float32) - m.wv.vectors_ngrams = np.full_like(m.wv.vectors_ngrams, 2, dtype=np.float32) mem = m._get_thread_working_mem() output = train_average_np(m, self.sentences, m.sv.vectors, mem) + self.assertEqual((4, 10), output) - self.assertTrue(np.allclose(1.0, m.sv[0])) - self.assertTrue(np.allclose(1.5, m.sv[2])) - self.assertTrue(np.allclose(2, m.sv[3])) + self.assertTrue((1.0 + EPS == m.sv[0]).all()) + self.assertTrue(np.allclose(368707.44, m.sv[2])) + self.assertTrue(np.allclose(961940.2, m.sv[3])) # "go" -> [1,1...] # oov: "12345" -> (14 hashes * 2) / 14 = 2 # (2 + 1) / 2 = 1.5 - def test_average_train_cy_ft(self): - ft = FastText(min_count=1, size=DIM) - ft.build_vocab(SENTENCES) - m = Average(ft) + def test_average_train_cy_ft_det(self): + m = Average(FT_DET) m.prep.prepare_vectors( sv=m.sv, total_sentences=len(self.sentences), update=False ) m._pre_train_calls() - m.wv.vectors = m.wv.vectors_vocab = np.ones_like(m.wv.vectors, dtype=np.float32) - m.wv.vectors_ngrams = np.full_like(m.wv.vectors_ngrams, 2, dtype=np.float32) mem = m._get_thread_working_mem() from fse.models.average_inner import train_average_cy @@ -120,11 +94,11 @@ def test_average_train_cy_ft(self): output = train_average_cy(m, self.sentences, m.sv.vectors, mem) self.assertEqual((4, 10), output) self.assertTrue((1.0 + EPS == m.sv[0]).all()) - self.assertTrue(np.allclose(1.5, m.sv[2])) - self.assertTrue(np.allclose(2, m.sv[3])) + self.assertTrue(np.allclose(368707.4, m.sv[2])) + self.assertTrue(np.allclose(961940., m.sv[3])) - def test_cy_equal_np_w2v(self): - m1 = Average(W2V) + def test_cy_equal_np_w2v_det(self): + m1 = Average(W2V_DET) m1.prep.prepare_vectors( sv=m1.sv, total_sentences=len(self.sentences), update=False ) @@ -132,7 +106,7 @@ def test_cy_equal_np_w2v(self): mem1 = m1._get_thread_working_mem() o1 = train_average_np(m1, self.sentences, m1.sv.vectors, mem1) - m2 = Average(W2V) + m2 = Average(W2V_DET) m2.prep.prepare_vectors( sv=m2.sv, total_sentences=len(self.sentences), update=False ) @@ -146,12 +120,8 @@ def test_cy_equal_np_w2v(self): self.assertEqual(o1, o2) self.assertTrue((m1.sv.vectors == m2.sv.vectors).all()) - def test_cy_equal_np_w2v_random(self): - w2v = Word2Vec(min_count=1, size=DIM) - # Random initialization - w2v.build_vocab(SENTENCES) - - m1 = Average(w2v) + def test_cy_equal_np_w2v_rng(self): + m1 = Average(W2V_RNG) m1.prep.prepare_vectors( sv=m1.sv, total_sentences=len(self.sentences), update=False ) @@ -159,7 +129,7 @@ def test_cy_equal_np_w2v_random(self): mem1 = m1._get_thread_working_mem() o1 = train_average_np(m1, self.sentences, m1.sv.vectors, mem1) - m2 = Average(w2v) + m2 = Average(W2V_RNG) m2.prep.prepare_vectors( sv=m2.sv, total_sentences=len(self.sentences), update=False ) @@ -172,11 +142,8 @@ def test_cy_equal_np_w2v_random(self): self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - def test_cy_equal_np_ft_random(self): - ft = FastText(size=20, min_count=1) - ft.build_vocab(SENTENCES) - - m1 = Average(ft) + def test_cy_equal_np_ft_rng(self): + m1 = Average(FT_RNG) m1.prep.prepare_vectors( sv=m1.sv, total_sentences=len(self.sentences), update=False ) @@ -188,7 +155,7 @@ def test_cy_equal_np_ft_random(self): mem1 = m1._get_thread_working_mem() o1 = train_average_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - m2 = Average(ft) + m2 = Average(FT_RNG) m2.prep.prepare_vectors( sv=m2.sv, total_sentences=len(self.sentences), update=False ) @@ -227,9 +194,11 @@ def test_train_single_from_disk(self): p_res = Path("fse/test/test_data/test_vecs.vectors") p_target = Path("fse/test/test_data/test_vecs_wv.vectors") - se1 = Average(W2V) + se1 = Average(W2V_DET) se2 = Average( - W2V, sv_mapfile_path=str(p.absolute()), wv_mapfile_path=str(p.absolute()) + W2V_DET, + sv_mapfile_path=str(p.absolute()), + wv_mapfile_path=str(p.absolute()), ) se1.train([(s, i) for i, s in enumerate(SENTENCES)]) se2.train([(s, i) for i, s in enumerate(SENTENCES)]) @@ -247,9 +216,9 @@ def test_train_multi_from_disk(self): p_res = Path("fse/test/test_data/test_vecs.vectors") p_target = Path("fse/test/test_data/test_vecs_wv.vectors") - se1 = Average(W2V, workers=2) + se1 = Average(W2V_DET, workers=2) se2 = Average( - W2V, + W2V_DET, workers=2, sv_mapfile_path=str(p.absolute()), wv_mapfile_path=str(p.absolute()), @@ -266,11 +235,23 @@ def test_train_multi_from_disk(self): p_target.unlink() def test_check_parameter_sanity(self): - se = Average(W2V) + se = Average(W2V_DET) se.word_weights = np.full(20, 2.0, dtype=np.float32) with self.assertRaises(ValueError): se._check_parameter_sanity() + se = Average(W2V_DET, window_size=0) + with self.assertRaises(ValueError): + se._check_parameter_sanity() + + se = Average(W2V_DET, window_size=3, window_stride=0) + with self.assertRaises(ValueError): + se._check_parameter_sanity() + + se = Average(W2V_DET, window_size=3, window_stride=4) + with self.assertRaises(ValueError): + se._check_parameter_sanity() + if __name__ == "__main__": logging.basicConfig( diff --git a/fse/test/test_base_s2v.py b/fse/test/test_base_s2v.py index cc50142..2915ea9 100644 --- a/fse/test/test_base_s2v.py +++ b/fse/test/test_base_s2v.py @@ -430,7 +430,7 @@ def temp_train_job(data_iterable, target, memory): report_delay=0.01, ) self.assertEqual((100, 200, 300), job_output) - + def test_memory_creation(self): se = BaseSentence2VecModel(W2V, workers=2) out = se._get_thread_working_mem() diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index 2001665..f77e0e3 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -45,8 +45,8 @@ FT_R = FastText(min_count=1, size=DIM) FT_R.build_vocab(SENTENCES) + class TestPoolingFunctions(unittest.TestCase): - def setUp(self): self.sentences = [ ["They", "admit"], @@ -69,6 +69,7 @@ def test_cython(self): MAX_NGRAMS_IN_BATCH, train_pooling_cy, ) + self.assertTrue(FAST_VERSION) self.assertTrue(callable(train_pooling_cy)) self.assertEqual(10000, MAX_WORDS_IN_BATCH) @@ -132,7 +133,7 @@ def test_pool_train_cy_w2v(self): mem = self.model._get_thread_working_mem() from fse.models.pooling_inner import train_pooling_cy - + output = train_pooling_cy( self.model, self.sentences, self.model.sv.vectors, mem ) @@ -141,410 +142,411 @@ def test_pool_train_cy_w2v(self): self.assertTrue((241 == self.model.sv[0]).all()) self.assertTrue((306 == self.model.sv[1]).all()) self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) - - def test_pool_train_np_ft(self): - m = MaxPooling(FT) - m.prep.prepare_vectors( - sv=m.sv, total_sentences=len(self.sentences), update=False - ) - m._pre_train_calls() - mem = m._get_thread_working_mem() - - output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) - self.assertEqual((5, 19), output) - self.assertTrue((241 == m.sv[0]).all()) - self.assertTrue( - np.allclose(737413.9, m.sv[2]) - ) - self.assertTrue( - np.allclose(1080970.2, m.sv[3]) - ) - - def test_pool_train_cy_ft(self): - m = MaxPooling(FT) - m.prep.prepare_vectors( - sv=m.sv, total_sentences=len(self.sentences), update=False - ) - m._pre_train_calls() - mem = m._get_thread_working_mem() - - from fse.models.pooling_inner import train_pooling_cy - - output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) - - self.assertEqual((5, 19), output) - self.assertTrue((241 == m.sv[0]).all()) - self.assertTrue( - np.allclose(737413.9, m.sv[2]) - ) - self.assertTrue( - np.allclose(1080970.2, m.sv[3]) - ) - - def test_pool_cy_equal_np_w2v(self): - m1 = MaxPooling(W2V) - m1.prep.prepare_vectors( - sv=m1.sv, total_sentences=len(self.sentences), update=False - ) - m1._pre_train_calls() - mem1 = m1._get_thread_working_mem() - o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) - - m2 = MaxPooling(W2V) - m2.prep.prepare_vectors( - sv=m2.sv, total_sentences=len(self.sentences), update=False - ) - m2._pre_train_calls() - mem2 = m2._get_thread_working_mem() - - from fse.models.pooling_inner import train_pooling_cy - - o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) - - self.assertEqual(o1, o2) - self.assertTrue((m1.sv.vectors == m2.sv.vectors).all()) - - def test_pool_cy_equal_np_w2v_random(self): - w2v = Word2Vec(min_count=1, size=DIM) - # Random initialization - w2v.build_vocab(SENTENCES) - - m1 = MaxPooling(w2v) - m1.prep.prepare_vectors( - sv=m1.sv, total_sentences=len(self.sentences), update=False - ) - m1._pre_train_calls() - mem1 = m1._get_thread_working_mem() - o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) - - m2 = MaxPooling(w2v) - m2.prep.prepare_vectors( - sv=m2.sv, total_sentences=len(self.sentences), update=False - ) - m2._pre_train_calls() - mem2 = m2._get_thread_working_mem() - - from fse.models.pooling_inner import train_pooling_cy - - o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) - - self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - - def test_pool_cy_equal_np_ft_random(self): - ft = FastText(size=20, min_count=1) - ft.build_vocab(SENTENCES) - - m1 = MaxPooling(ft) - m1.prep.prepare_vectors( - sv=m1.sv, total_sentences=len(self.sentences), update=False - ) - m1._pre_train_calls() - - from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH - - m1.batch_ngrams = MAX_NGRAMS_IN_BATCH - mem1 = m1._get_thread_working_mem() - o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - - m2 = MaxPooling(ft) - m2.prep.prepare_vectors( - sv=m2.sv, total_sentences=len(self.sentences), update=False - ) - m2._pre_train_calls() - mem2 = m2._get_thread_working_mem() - - from fse.models.pooling_inner import train_pooling_cy - - o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) - - self.assertEqual(o1, o2) - self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - - def test_pool_np_w2v_non_negative(self): - mpool = MaxPooling(W2V_R) - mpool.train(self.sentences) - self.assertTrue((mpool.sv.vectors >= 0).all()) - - def test_pool_np_ft_non_negative(self): - mpool = MaxPooling(FT_R) - mpool.train(self.sentences) - self.assertTrue((mpool.sv.vectors >= 0).all()) + # def test_pool_train_np_ft(self): + # m = MaxPooling(FT) + # m.prep.prepare_vectors( + # sv=m.sv, total_sentences=len(self.sentences), update=False + # ) + # m._pre_train_calls() + # mem = m._get_thread_working_mem() + + # output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) + + # self.assertEqual((5, 19), output) + # self.assertTrue((241 == m.sv[0]).all()) + # self.assertTrue( + # np.allclose(737413.9, m.sv[2]) + # ) + # self.assertTrue( + # np.allclose(1080970.2, m.sv[3]) + # ) + + # def test_pool_train_cy_ft(self): + # m = MaxPooling(FT) + # m.prep.prepare_vectors( + # sv=m.sv, total_sentences=len(self.sentences), update=False + # ) + # m._pre_train_calls() + # mem = m._get_thread_working_mem() + + # from fse.models.pooling_inner import train_pooling_cy + + # output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) + + # self.assertEqual((5, 19), output) + # self.assertTrue((241 == m.sv[0]).all()) + # self.assertTrue( + # np.allclose(737413.9, m.sv[2]) + # ) + # self.assertTrue( + # np.allclose(1080970.2, m.sv[3]) + # ) + + # def test_pool_cy_equal_np_w2v(self): + # m1 = MaxPooling(W2V) + # m1.prep.prepare_vectors( + # sv=m1.sv, total_sentences=len(self.sentences), update=False + # ) + # m1._pre_train_calls() + # mem1 = m1._get_thread_working_mem() + # o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + + # m2 = MaxPooling(W2V) + # m2.prep.prepare_vectors( + # sv=m2.sv, total_sentences=len(self.sentences), update=False + # ) + # m2._pre_train_calls() + # mem2 = m2._get_thread_working_mem() + + # from fse.models.pooling_inner import train_pooling_cy + + # o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + + # self.assertEqual(o1, o2) + # self.assertTrue((m1.sv.vectors == m2.sv.vectors).all()) + + # def test_pool_cy_equal_np_w2v_random(self): + # w2v = Word2Vec(min_count=1, size=DIM) + # # Random initialization + # w2v.build_vocab(SENTENCES) + + # m1 = MaxPooling(w2v) + # m1.prep.prepare_vectors( + # sv=m1.sv, total_sentences=len(self.sentences), update=False + # ) + # m1._pre_train_calls() + # mem1 = m1._get_thread_working_mem() + # o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + + # m2 = MaxPooling(w2v) + # m2.prep.prepare_vectors( + # sv=m2.sv, total_sentences=len(self.sentences), update=False + # ) + # m2._pre_train_calls() + # mem2 = m2._get_thread_working_mem() + + # from fse.models.pooling_inner import train_pooling_cy + + # o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + + # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + # def test_pool_cy_equal_np_ft_random(self): + # ft = FastText(size=20, min_count=1) + # ft.build_vocab(SENTENCES) + + # m1 = MaxPooling(ft) + # m1.prep.prepare_vectors( + # sv=m1.sv, total_sentences=len(self.sentences), update=False + # ) + # m1._pre_train_calls() + + # from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH + + # m1.batch_ngrams = MAX_NGRAMS_IN_BATCH + # mem1 = m1._get_thread_working_mem() + # o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) + + # m2 = MaxPooling(ft) + # m2.prep.prepare_vectors( + # sv=m2.sv, total_sentences=len(self.sentences), update=False + # ) + # m2._pre_train_calls() + # mem2 = m2._get_thread_working_mem() + + # from fse.models.pooling_inner import train_pooling_cy + + # o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) + + # self.assertEqual(o1, o2) + # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + # def test_pool_np_w2v_non_negative(self): + # mpool = MaxPooling(W2V_R) + # mpool.train(self.sentences) + # self.assertTrue((mpool.sv.vectors >= 0).all()) + + # def test_pool_np_ft_non_negative(self): + # mpool = MaxPooling(FT_R) + # mpool.train(self.sentences) + # self.assertTrue((mpool.sv.vectors >= 0).all()) ### Hierarchical Tests start here - def test_hier_pool_train_np_w2v(self): - self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) - mem = self.model._get_thread_working_mem() - - self.model.hierarchical = True - - output = train_pooling_np( - self.model, self.sentences, self.model.sv.vectors, mem - ) - self.model.hierarchical = False + # def test_hier_pool_train_np_w2v(self): + # self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + # mem = self.model._get_thread_working_mem() + + # self.model.hierarchical = True + + # output = train_pooling_np( + # self.model, self.sentences, self.model.sv.vectors, mem + # ) + # self.model.hierarchical = False + + # self.assertEqual((5, 14), output) + # self.assertTrue((183 == self.model.sv[0]).all()) + # self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) + + # def test_hier_pool_train_cy_w2v(self): + # self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + # mem = self.model._get_thread_working_mem() + + # self.model.hierarchical = True + + # from fse.models.pooling_inner import train_pooling_cy + + # output = train_pooling_cy( + # self.model, self.sentences, self.model.sv.vectors, mem + # ) + # self.model.hierarchical = False + + # self.assertEqual((5, 14), output) + # self.assertTrue((183 == self.model.sv[0]).all()) + # self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) + + # def test_hier_pool_train_np_ft(self): + # m = MaxPooling(FT, hierarchical=True) + # m.prep.prepare_vectors( + # sv=m.sv, total_sentences=len(self.sentences), update=False + # ) + # m._pre_train_calls() + # mem = m._get_thread_working_mem() + + # output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) + + # self.assertEqual((5, 19), output) + # self.assertTrue((183 == m.sv[0]).all()) + # self.assertTrue(np.allclose(737413.9, m.sv[2])) + # self.assertTrue(np.allclose(1080970.2, m.sv[3])) + # """ + # Note to future self: + # Due to the size of the ngram vectors, + # an ngram at the last position of the senence + # will always be the highest value. + # TODO: This unittest is thus a bit flawed. Maybe fix? + # """ + + # def test_hier_pool_train_cy_ft(self): + # m = MaxPooling(FT, hierarchical=True) + # m.prep.prepare_vectors( + # sv=m.sv, total_sentences=len(self.sentences), update=False + # ) + # m._pre_train_calls() + # mem = m._get_thread_working_mem() + + # from fse.models.pooling_inner import train_pooling_cy + + # output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) - self.assertEqual((5, 14), output) - self.assertTrue((183 == self.model.sv[0]).all()) - self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) + # self.assertEqual((5, 19), output) + # self.assertTrue((183 == m.sv[0]).all()) + # self.assertTrue(np.allclose(737413.9, m.sv[2])) + # self.assertTrue(np.allclose(1080970.2, m.sv[3])) + + # def test_hier_pool_cy_equal_np_w2v_random(self): + # w2v = Word2Vec(min_count=1, size=DIM) + # # Random initialization + # w2v.build_vocab(SENTENCES) + + # m1 = MaxPooling(w2v, hierarchical=True) + # m1.prep.prepare_vectors( + # sv=m1.sv, total_sentences=len(self.sentences), update=False + # ) + # m1._pre_train_calls() + # mem1 = m1._get_thread_working_mem() + # o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + + # m2 = MaxPooling(w2v, hierarchical=True) + # m2.prep.prepare_vectors( + # sv=m2.sv, total_sentences=len(self.sentences), update=False + # ) + # m2._pre_train_calls() + # mem2 = m2._get_thread_working_mem() + + # from fse.models.pooling_inner import train_pooling_cy + + # o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + + # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + # def test_hier_pool_cy_equal_np_ft_random(self): + # ft = FastText(size=20, min_count=1) + # ft.build_vocab(SENTENCES) + + # m1 = MaxPooling(ft, hierarchical=True) + # m1.prep.prepare_vectors( + # sv=m1.sv, total_sentences=len(self.sentences), update=False + # ) + # m1._pre_train_calls() + + # from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH + + # m1.batch_ngrams = MAX_NGRAMS_IN_BATCH + # mem1 = m1._get_thread_working_mem() + # o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) + + # m2 = MaxPooling(ft, hierarchical=True) + # m2.prep.prepare_vectors( + # sv=m2.sv, total_sentences=len(self.sentences), update=False + # ) + # m2._pre_train_calls() + # mem2 = m2._get_thread_working_mem() + + # from fse.models.pooling_inner import train_pooling_cy + + # o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) + + # self.assertEqual(o1, o2) + # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + # def test_hier_pool_cy_w2v_non_negative(self): + # mpool = MaxPooling(W2V_R, hierarchical=True) + # mpool.train(self.sentences) + # self.assertTrue((mpool.sv.vectors >= 0).all()) + + # def test_hier_pool_cy_ft_non_negative(self): + # mpool = MaxPooling(FT_R, hierarchical=True) + # mpool.train(self.sentences) + # self.assertTrue((mpool.sv.vectors >= 0).all()) + + # ### Hierarchical Test + Stride start here + + # def test_hier_pool_stride_train_np_w2v(self): + # self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + # mem = self.model._get_thread_working_mem() + + # self.model.hierarchical = True + # self.model.window_stride = 5 + + # output = train_pooling_np( + # self.model, self.sentences, self.model.sv.vectors, mem + # ) + # self.model.hierarchical = False + # self.model.window_stride = 1 + + # self.assertEqual((5, 14), output) + # self.assertTrue((183 == self.model.sv[0]).all()) + # self.assertTrue((231 == self.model.sv[4]).all()) + + # def test_hier_pool_stride_train_cy_w2v(self): + # self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + # mem = self.model._get_thread_working_mem() - def test_hier_pool_train_cy_w2v(self): - self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) - mem = self.model._get_thread_working_mem() + # self.model.hierarchical = True + # self.model.window_stride = 5 - self.model.hierarchical = True + # from fse.models.pooling_inner import train_pooling_cy + + # output = train_pooling_cy( + # self.model, self.sentences, self.model.sv.vectors, mem + # ) + # self.model.hierarchical = False + # self.model.window_stride = 1 + + # self.assertEqual((5, 14), output) + # self.assertTrue((183 == self.model.sv[0]).all()) + # self.assertTrue((231 == self.model.sv[4]).all()) - from fse.models.pooling_inner import train_pooling_cy + # def test_hier_pool_stride_train_np_ft(self): + # m = MaxPooling(FT, hierarchical=True, window_stride=3) + # m.prep.prepare_vectors( + # sv=m.sv, total_sentences=len(self.sentences), update=False + # ) + # m._pre_train_calls() + # mem = m._get_thread_working_mem() - output = train_pooling_cy( - self.model, self.sentences, self.model.sv.vectors, mem - ) - self.model.hierarchical = False + # output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) - self.assertEqual((5, 14), output) - self.assertTrue((183 == self.model.sv[0]).all()) - self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) + # self.assertEqual((5, 19), output) + # self.assertTrue((183 == m.sv[0]).all()) + # self.assertTrue(np.allclose(368871.94, m.sv[2])) + # self.assertTrue(np.allclose(961940.2, m.sv[3])) - def test_hier_pool_train_np_ft(self): - m = MaxPooling(FT, hierarchical=True) - m.prep.prepare_vectors( - sv=m.sv, total_sentences=len(self.sentences), update=False - ) - m._pre_train_calls() - mem = m._get_thread_working_mem() - - output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) - - self.assertEqual((5, 19), output) - self.assertTrue((183 == m.sv[0]).all()) - self.assertTrue(np.allclose(737413.9, m.sv[2])) - self.assertTrue(np.allclose(1080970.2, m.sv[3])) - """ - Note to future self: - Due to the size of the ngram vectors, - an ngram at the last position of the senence - will always be the highest value. - TODO: This unittest is thus a bit flawed. Maybe fix? - """ - - def test_hier_pool_train_cy_ft(self): - m = MaxPooling(FT, hierarchical=True) - m.prep.prepare_vectors( - sv=m.sv, total_sentences=len(self.sentences), update=False - ) - m._pre_train_calls() - mem = m._get_thread_working_mem() + # def test_hier_pool_stride_train_cy_ft(self): + # m = MaxPooling(FT, hierarchical=True, window_stride=3) + # m.prep.prepare_vectors( + # sv=m.sv, total_sentences=len(self.sentences), update=False + # ) + # m._pre_train_calls() + # mem = m._get_thread_working_mem() + + # from fse.models.pooling_inner import train_pooling_cy - from fse.models.pooling_inner import train_pooling_cy + # output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) + + # self.assertEqual((5, 19), output) + # self.assertTrue((183 == m.sv[0]).all()) + # self.assertTrue(np.allclose(368871.94, m.sv[2])) + # self.assertTrue(np.allclose(961940.2, m.sv[3])) - output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) + # def test_hier_pool_stride_cy_equal_np_w2v_random(self): + # w2v = Word2Vec(min_count=1, size=DIM) + # # Random initialization + # w2v.build_vocab(SENTENCES) - self.assertEqual((5, 19), output) - self.assertTrue((183 == m.sv[0]).all()) - self.assertTrue(np.allclose(737413.9, m.sv[2])) - self.assertTrue(np.allclose(1080970.2, m.sv[3])) + # m1 = MaxPooling(w2v, hierarchical=True, window_stride=4) + # m1.prep.prepare_vectors( + # sv=m1.sv, total_sentences=len(self.sentences), update=False + # ) + # m1._pre_train_calls() + # mem1 = m1._get_thread_working_mem() + # o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) - def test_hier_pool_cy_equal_np_w2v_random(self): - w2v = Word2Vec(min_count=1, size=DIM) - # Random initialization - w2v.build_vocab(SENTENCES) + # m2 = MaxPooling(w2v, hierarchical=True, window_stride=4) + # m2.prep.prepare_vectors( + # sv=m2.sv, total_sentences=len(self.sentences), update=False + # ) + # m2._pre_train_calls() + # mem2 = m2._get_thread_working_mem() - m1 = MaxPooling(w2v, hierarchical=True) - m1.prep.prepare_vectors( - sv=m1.sv, total_sentences=len(self.sentences), update=False - ) - m1._pre_train_calls() - mem1 = m1._get_thread_working_mem() - o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + # from fse.models.pooling_inner import train_pooling_cy + + # o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) - m2 = MaxPooling(w2v, hierarchical=True) - m2.prep.prepare_vectors( - sv=m2.sv, total_sentences=len(self.sentences), update=False - ) - m2._pre_train_calls() - mem2 = m2._get_thread_working_mem() + # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + # def test_hier_pool_stride_cy_equal_np_ft_random(self): + # ft = FastText(size=20, min_count=1) + # ft.build_vocab(SENTENCES) - from fse.models.pooling_inner import train_pooling_cy + # m1 = MaxPooling(ft, hierarchical=True, window_stride=5) + # m1.prep.prepare_vectors( + # sv=m1.sv, total_sentences=len(self.sentences), update=False + # ) + # m1._pre_train_calls() - o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) - - self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - - def test_hier_pool_cy_equal_np_ft_random(self): - ft = FastText(size=20, min_count=1) - ft.build_vocab(SENTENCES) - - m1 = MaxPooling(ft, hierarchical=True) - m1.prep.prepare_vectors( - sv=m1.sv, total_sentences=len(self.sentences), update=False - ) - m1._pre_train_calls() - - from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH - - m1.batch_ngrams = MAX_NGRAMS_IN_BATCH - mem1 = m1._get_thread_working_mem() - o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - - m2 = MaxPooling(ft, hierarchical=True) - m2.prep.prepare_vectors( - sv=m2.sv, total_sentences=len(self.sentences), update=False - ) - m2._pre_train_calls() - mem2 = m2._get_thread_working_mem() - - from fse.models.pooling_inner import train_pooling_cy - - o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) - - self.assertEqual(o1, o2) - self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - - def test_hier_pool_cy_w2v_non_negative(self): - mpool = MaxPooling(W2V_R, hierarchical=True) - mpool.train(self.sentences) - self.assertTrue((mpool.sv.vectors >= 0).all()) - - def test_hier_pool_cy_ft_non_negative(self): - mpool = MaxPooling(FT_R, hierarchical=True) - mpool.train(self.sentences) - self.assertTrue((mpool.sv.vectors >= 0).all()) - - ### Hierarchical Test + Stride start here - - def test_hier_pool_stride_train_np_w2v(self): - self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) - mem = self.model._get_thread_working_mem() - - self.model.hierarchical = True - self.model.window_stride = 5 - - output = train_pooling_np( - self.model, self.sentences, self.model.sv.vectors, mem - ) - self.model.hierarchical = False - self.model.window_stride = 1 - - self.assertEqual((5, 14), output) - self.assertTrue((183 == self.model.sv[0]).all()) - self.assertTrue((231 == self.model.sv[4]).all()) - - def test_hier_pool_stride_train_cy_w2v(self): - self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) - mem = self.model._get_thread_working_mem() - - self.model.hierarchical = True - self.model.window_stride = 5 - - from fse.models.pooling_inner import train_pooling_cy - - output = train_pooling_cy( - self.model, self.sentences, self.model.sv.vectors, mem - ) - self.model.hierarchical = False - self.model.window_stride = 1 - - self.assertEqual((5, 14), output) - self.assertTrue((183 == self.model.sv[0]).all()) - self.assertTrue((231 == self.model.sv[4]).all()) - - def test_hier_pool_stride_train_np_ft(self): - m = MaxPooling(FT, hierarchical=True, window_stride=3) - m.prep.prepare_vectors( - sv=m.sv, total_sentences=len(self.sentences), update=False - ) - m._pre_train_calls() - mem = m._get_thread_working_mem() - - output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) - - self.assertEqual((5, 19), output) - self.assertTrue((183 == m.sv[0]).all()) - self.assertTrue(np.allclose(368871.94, m.sv[2])) - self.assertTrue(np.allclose(961940.2, m.sv[3])) - - def test_hier_pool_stride_train_cy_ft(self): - m = MaxPooling(FT, hierarchical=True, window_stride=3) - m.prep.prepare_vectors( - sv=m.sv, total_sentences=len(self.sentences), update=False - ) - m._pre_train_calls() - mem = m._get_thread_working_mem() - - from fse.models.pooling_inner import train_pooling_cy - - output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) - - self.assertEqual((5, 19), output) - self.assertTrue((183 == m.sv[0]).all()) - self.assertTrue(np.allclose(368871.94, m.sv[2])) - self.assertTrue(np.allclose(961940.2, m.sv[3])) - - def test_hier_pool_stride_cy_equal_np_w2v_random(self): - w2v = Word2Vec(min_count=1, size=DIM) - # Random initialization - w2v.build_vocab(SENTENCES) - - m1 = MaxPooling(w2v, hierarchical=True, window_stride=4) - m1.prep.prepare_vectors( - sv=m1.sv, total_sentences=len(self.sentences), update=False - ) - m1._pre_train_calls() - mem1 = m1._get_thread_working_mem() - o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) - - m2 = MaxPooling(w2v, hierarchical=True, window_stride=4) - m2.prep.prepare_vectors( - sv=m2.sv, total_sentences=len(self.sentences), update=False - ) - m2._pre_train_calls() - mem2 = m2._get_thread_working_mem() - - from fse.models.pooling_inner import train_pooling_cy - - o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) - - self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - - def test_hier_pool_stride_cy_equal_np_ft_random(self): - ft = FastText(size=20, min_count=1) - ft.build_vocab(SENTENCES) - - m1 = MaxPooling(ft, hierarchical=True, window_stride=5) - m1.prep.prepare_vectors( - sv=m1.sv, total_sentences=len(self.sentences), update=False - ) - m1._pre_train_calls() - - from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH - - m1.batch_ngrams = MAX_NGRAMS_IN_BATCH - mem1 = m1._get_thread_working_mem() - o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - - m2 = MaxPooling(ft, hierarchical=True, window_stride=5) - m2.prep.prepare_vectors( - sv=m2.sv, total_sentences=len(self.sentences), update=False - ) - m2._pre_train_calls() - mem2 = m2._get_thread_working_mem() - - from fse.models.pooling_inner import train_pooling_cy + # from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH + + # m1.batch_ngrams = MAX_NGRAMS_IN_BATCH + # mem1 = m1._get_thread_working_mem() + # o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) + + # m2 = MaxPooling(ft, hierarchical=True, window_stride=5) + # m2.prep.prepare_vectors( + # sv=m2.sv, total_sentences=len(self.sentences), update=False + # ) + # m2._pre_train_calls() + # mem2 = m2._get_thread_working_mem() - o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) + # from fse.models.pooling_inner import train_pooling_cy + + # o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) - self.assertEqual(o1, o2) - self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + # self.assertEqual(o1, o2) + # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - def test_hier_pool_stride_np_w2v_non_negative(self): - mpool = MaxPooling(W2V_R, hierarchical=True, window_stride=4) - mpool.train(self.sentences) - self.assertTrue((mpool.sv.vectors >= 0).all()) + # def test_hier_pool_stride_np_w2v_non_negative(self): + # mpool = MaxPooling(W2V_R, hierarchical=True, window_stride=4) + # mpool.train(self.sentences) + # self.assertTrue((mpool.sv.vectors >= 0).all()) + + # def test_hier_pool_stride_np_ft_non_negative(self): + # mpool = MaxPooling(FT_R, hierarchical=True, window_stride=4) + # mpool.train(self.sentences) + # self.assertTrue((mpool.sv.vectors >= 0).all()) - def test_hier_pool_stride_np_ft_non_negative(self): - mpool = MaxPooling(FT_R, hierarchical=True, window_stride=4) - mpool.train(self.sentences) - self.assertTrue((mpool.sv.vectors >= 0).all()) if __name__ == "__main__": logging.basicConfig( diff --git a/fse/test/test_utils.py b/fse/test/test_utils.py index 550e540..4ef3c62 100644 --- a/fse/test/test_utils.py +++ b/fse/test/test_utils.py @@ -4,17 +4,16 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -import logging -import unittest - -import numpy as np - -from fse.models.utils import compute_principal_components, remove_principal_components - -logger = logging.getLogger(__name__) +from shared_imports import * +from fse.models.utils import ( + compute_principal_components, + remove_principal_components, + get_ft_word_vector, +) class TestUtils(unittest.TestCase): + def test_compute_components(self): m = np.random.uniform(size=(500, 10)).astype(np.float32) out = compute_principal_components(vectors=m) @@ -73,6 +72,15 @@ def test_madvise(self): ) p.unlink() + def test_ft_word_vector(self): + from fse.models.average import Average + model = Average(FT_DET) + + word = "123456789" + vec = get_ft_word_vector(word, model) + + self.assertIsInstance(vec, np.ndarray) + self.assertTrue(np.allclose(max(vec), 979980.94)) if __name__ == "__main__": logging.basicConfig( From 76498200c5d6b4ac7c0682a71d74ca32c0c539f1 Mon Sep 17 00:00:00 2001 From: OliverB Date: Fri, 22 May 2020 13:49:48 +0200 Subject: [PATCH 45/63] black --- fse/test/shared_imports.py | 6 +- fse/test/test_average.py | 3 +- fse/test/test_pooling.py | 889 +++++++++++++++++++------------------ fse/test/test_utils.py | 6 +- 4 files changed, 455 insertions(+), 449 deletions(-) diff --git a/fse/test/shared_imports.py b/fse/test/shared_imports.py index 3578c5a..942f402 100644 --- a/fse/test/shared_imports.py +++ b/fse/test/shared_imports.py @@ -34,9 +34,9 @@ FT_DET.wv.vectors = FT_DET.wv.vectors_vocab = np.ones_like( FT_DET.wv.vectors, dtype=np.float32 ) -FT_DET.wv.vectors_ngrams[:,] = np.arange(len(FT_DET.wv.vectors_ngrams), dtype=np.float32)[ - :, None -] +FT_DET.wv.vectors_ngrams[:,] = np.arange( + len(FT_DET.wv.vectors_ngrams), dtype=np.float32 +)[:, None] # Random FT FT_RNG = FastText(min_count=1, size=DIM) diff --git a/fse/test/test_average.py b/fse/test/test_average.py index d95c108..10c719a 100644 --- a/fse/test/test_average.py +++ b/fse/test/test_average.py @@ -13,6 +13,7 @@ from fse.models.average import Average, train_average_np + class TestAverageFunctions(unittest.TestCase): def setUp(self): self.sentences = [ @@ -95,7 +96,7 @@ def test_average_train_cy_ft_det(self): self.assertEqual((4, 10), output) self.assertTrue((1.0 + EPS == m.sv[0]).all()) self.assertTrue(np.allclose(368707.4, m.sv[2])) - self.assertTrue(np.allclose(961940., m.sv[3])) + self.assertTrue(np.allclose(961940.0, m.sv[3])) def test_cy_equal_np_w2v_det(self): m1 = Average(W2V_DET) diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index f77e0e3..050fbeb 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -62,490 +62,493 @@ def setUp(self): ) self.model._pre_train_calls() - def test_cython(self): - from fse.models.pooling_inner import ( - FAST_VERSION, - MAX_WORDS_IN_BATCH, - MAX_NGRAMS_IN_BATCH, - train_pooling_cy, - ) - - self.assertTrue(FAST_VERSION) - self.assertTrue(callable(train_pooling_cy)) - self.assertEqual(10000, MAX_WORDS_IN_BATCH) - self.assertEqual(40, MAX_NGRAMS_IN_BATCH) - - def test_check_parameter_sanity(self): - se = MaxPooling(W2V) - se.word_weights = np.full(20, 2.0, dtype=np.float32) - with self.assertRaises(ValueError): - se._check_parameter_sanity() - - se = MaxPooling(W2V, window_size=0) - with self.assertRaises(ValueError): - se._check_parameter_sanity() - - se = MaxPooling(W2V, window_size=3, window_stride=0) - with self.assertRaises(ValueError): - se._check_parameter_sanity() - - se = MaxPooling(W2V, window_size=3, window_stride=4) - with self.assertRaises(ValueError): - se._check_parameter_sanity() - - def test_train(self): - self.assertEqual( - (100, 1450), self.model.train([(s, i) for i, s in enumerate(SENTENCES)]) - ) - - def test_do_train_job(self): - self.model.prep.prepare_vectors( - sv=self.model.sv, total_sentences=len(SENTENCES), update=True - ) - mem = self.model._get_thread_working_mem() - self.assertEqual( - (100, 1450), - self.model._do_train_job( - [(s, i) for i, s in enumerate(SENTENCES)], - target=self.model.sv.vectors, - memory=mem, - ), - ) - self.assertEqual((105, DIM), self.model.sv.vectors.shape) - - ### Basic Pooling Tests start here - - def test_pool_train_np_w2v(self): - self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) - mem = self.model._get_thread_working_mem() - - output = train_pooling_np( - self.model, self.sentences, self.model.sv.vectors, mem - ) - - self.assertEqual((5, 14), output) - self.assertTrue((241 == self.model.sv[0]).all()) - self.assertTrue((306 == self.model.sv[1]).all()) - self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) - - def test_pool_train_cy_w2v(self): - self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) - mem = self.model._get_thread_working_mem() - - from fse.models.pooling_inner import train_pooling_cy - - output = train_pooling_cy( - self.model, self.sentences, self.model.sv.vectors, mem - ) - - self.assertEqual((5, 14), output) - self.assertTrue((241 == self.model.sv[0]).all()) - self.assertTrue((306 == self.model.sv[1]).all()) - self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) - - # def test_pool_train_np_ft(self): - # m = MaxPooling(FT) - # m.prep.prepare_vectors( - # sv=m.sv, total_sentences=len(self.sentences), update=False - # ) - # m._pre_train_calls() - # mem = m._get_thread_working_mem() - - # output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) - - # self.assertEqual((5, 19), output) - # self.assertTrue((241 == m.sv[0]).all()) - # self.assertTrue( - # np.allclose(737413.9, m.sv[2]) - # ) - # self.assertTrue( - # np.allclose(1080970.2, m.sv[3]) - # ) - - # def test_pool_train_cy_ft(self): - # m = MaxPooling(FT) - # m.prep.prepare_vectors( - # sv=m.sv, total_sentences=len(self.sentences), update=False - # ) - # m._pre_train_calls() - # mem = m._get_thread_working_mem() - - # from fse.models.pooling_inner import train_pooling_cy - - # output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) - - # self.assertEqual((5, 19), output) - # self.assertTrue((241 == m.sv[0]).all()) - # self.assertTrue( - # np.allclose(737413.9, m.sv[2]) - # ) - # self.assertTrue( - # np.allclose(1080970.2, m.sv[3]) + # def test_cython(self): + # from fse.models.pooling_inner import ( + # FAST_VERSION, + # MAX_WORDS_IN_BATCH, + # MAX_NGRAMS_IN_BATCH, + # train_pooling_cy, # ) - # def test_pool_cy_equal_np_w2v(self): - # m1 = MaxPooling(W2V) - # m1.prep.prepare_vectors( - # sv=m1.sv, total_sentences=len(self.sentences), update=False - # ) - # m1._pre_train_calls() - # mem1 = m1._get_thread_working_mem() - # o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + # self.assertTrue(FAST_VERSION) + # self.assertTrue(callable(train_pooling_cy)) + # self.assertEqual(10000, MAX_WORDS_IN_BATCH) + # self.assertEqual(40, MAX_NGRAMS_IN_BATCH) - # m2 = MaxPooling(W2V) - # m2.prep.prepare_vectors( - # sv=m2.sv, total_sentences=len(self.sentences), update=False - # ) - # m2._pre_train_calls() - # mem2 = m2._get_thread_working_mem() + # def test_check_parameter_sanity(self): + # se = MaxPooling(W2V) + # se.word_weights = np.full(20, 2.0, dtype=np.float32) + # with self.assertRaises(ValueError): + # se._check_parameter_sanity() - # from fse.models.pooling_inner import train_pooling_cy + # se = MaxPooling(W2V, window_size=0) + # with self.assertRaises(ValueError): + # se._check_parameter_sanity() - # o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + # se = MaxPooling(W2V, window_size=3, window_stride=0) + # with self.assertRaises(ValueError): + # se._check_parameter_sanity() - # self.assertEqual(o1, o2) - # self.assertTrue((m1.sv.vectors == m2.sv.vectors).all()) + # se = MaxPooling(W2V, window_size=3, window_stride=4) + # with self.assertRaises(ValueError): + # se._check_parameter_sanity() - # def test_pool_cy_equal_np_w2v_random(self): - # w2v = Word2Vec(min_count=1, size=DIM) - # # Random initialization - # w2v.build_vocab(SENTENCES) - - # m1 = MaxPooling(w2v) - # m1.prep.prepare_vectors( - # sv=m1.sv, total_sentences=len(self.sentences), update=False - # ) - # m1._pre_train_calls() - # mem1 = m1._get_thread_working_mem() - # o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) - - # m2 = MaxPooling(w2v) - # m2.prep.prepare_vectors( - # sv=m2.sv, total_sentences=len(self.sentences), update=False + # def test_train(self): + # self.assertEqual( + # (100, 1450), self.model.train([(s, i) for i, s in enumerate(SENTENCES)]) # ) - # m2._pre_train_calls() - # mem2 = m2._get_thread_working_mem() - - # from fse.models.pooling_inner import train_pooling_cy - - # o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) - - # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - - # def test_pool_cy_equal_np_ft_random(self): - # ft = FastText(size=20, min_count=1) - # ft.build_vocab(SENTENCES) - - # m1 = MaxPooling(ft) - # m1.prep.prepare_vectors( - # sv=m1.sv, total_sentences=len(self.sentences), update=False - # ) - # m1._pre_train_calls() - - # from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH - - # m1.batch_ngrams = MAX_NGRAMS_IN_BATCH - # mem1 = m1._get_thread_working_mem() - # o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - # m2 = MaxPooling(ft) - # m2.prep.prepare_vectors( - # sv=m2.sv, total_sentences=len(self.sentences), update=False + # def test_do_train_job(self): + # self.model.prep.prepare_vectors( + # sv=self.model.sv, total_sentences=len(SENTENCES), update=True # ) - # m2._pre_train_calls() - # mem2 = m2._get_thread_working_mem() - - # from fse.models.pooling_inner import train_pooling_cy - - # o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) - - # self.assertEqual(o1, o2) - # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - - # def test_pool_np_w2v_non_negative(self): - # mpool = MaxPooling(W2V_R) - # mpool.train(self.sentences) - # self.assertTrue((mpool.sv.vectors >= 0).all()) - - # def test_pool_np_ft_non_negative(self): - # mpool = MaxPooling(FT_R) - # mpool.train(self.sentences) - # self.assertTrue((mpool.sv.vectors >= 0).all()) - - ### Hierarchical Tests start here - - # def test_hier_pool_train_np_w2v(self): - # self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) - # mem = self.model._get_thread_working_mem() - - # self.model.hierarchical = True - - # output = train_pooling_np( - # self.model, self.sentences, self.model.sv.vectors, mem - # ) - # self.model.hierarchical = False - - # self.assertEqual((5, 14), output) - # self.assertTrue((183 == self.model.sv[0]).all()) - # self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) - - # def test_hier_pool_train_cy_w2v(self): - # self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) # mem = self.model._get_thread_working_mem() - - # self.model.hierarchical = True - - # from fse.models.pooling_inner import train_pooling_cy - - # output = train_pooling_cy( - # self.model, self.sentences, self.model.sv.vectors, mem - # ) - # self.model.hierarchical = False - - # self.assertEqual((5, 14), output) - # self.assertTrue((183 == self.model.sv[0]).all()) - # self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) - - # def test_hier_pool_train_np_ft(self): - # m = MaxPooling(FT, hierarchical=True) - # m.prep.prepare_vectors( - # sv=m.sv, total_sentences=len(self.sentences), update=False - # ) - # m._pre_train_calls() - # mem = m._get_thread_working_mem() - - # output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) - - # self.assertEqual((5, 19), output) - # self.assertTrue((183 == m.sv[0]).all()) - # self.assertTrue(np.allclose(737413.9, m.sv[2])) - # self.assertTrue(np.allclose(1080970.2, m.sv[3])) - # """ - # Note to future self: - # Due to the size of the ngram vectors, - # an ngram at the last position of the senence - # will always be the highest value. - # TODO: This unittest is thus a bit flawed. Maybe fix? - # """ - - # def test_hier_pool_train_cy_ft(self): - # m = MaxPooling(FT, hierarchical=True) - # m.prep.prepare_vectors( - # sv=m.sv, total_sentences=len(self.sentences), update=False - # ) - # m._pre_train_calls() - # mem = m._get_thread_working_mem() - - # from fse.models.pooling_inner import train_pooling_cy - - # output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) - - # self.assertEqual((5, 19), output) - # self.assertTrue((183 == m.sv[0]).all()) - # self.assertTrue(np.allclose(737413.9, m.sv[2])) - # self.assertTrue(np.allclose(1080970.2, m.sv[3])) - - # def test_hier_pool_cy_equal_np_w2v_random(self): - # w2v = Word2Vec(min_count=1, size=DIM) - # # Random initialization - # w2v.build_vocab(SENTENCES) - - # m1 = MaxPooling(w2v, hierarchical=True) - # m1.prep.prepare_vectors( - # sv=m1.sv, total_sentences=len(self.sentences), update=False - # ) - # m1._pre_train_calls() - # mem1 = m1._get_thread_working_mem() - # o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) - - # m2 = MaxPooling(w2v, hierarchical=True) - # m2.prep.prepare_vectors( - # sv=m2.sv, total_sentences=len(self.sentences), update=False - # ) - # m2._pre_train_calls() - # mem2 = m2._get_thread_working_mem() - - # from fse.models.pooling_inner import train_pooling_cy - - # o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) - - # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - - # def test_hier_pool_cy_equal_np_ft_random(self): - # ft = FastText(size=20, min_count=1) - # ft.build_vocab(SENTENCES) - - # m1 = MaxPooling(ft, hierarchical=True) - # m1.prep.prepare_vectors( - # sv=m1.sv, total_sentences=len(self.sentences), update=False - # ) - # m1._pre_train_calls() - - # from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH - - # m1.batch_ngrams = MAX_NGRAMS_IN_BATCH - # mem1 = m1._get_thread_working_mem() - # o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - - # m2 = MaxPooling(ft, hierarchical=True) - # m2.prep.prepare_vectors( - # sv=m2.sv, total_sentences=len(self.sentences), update=False + # self.assertEqual( + # (100, 1450), + # self.model._do_train_job( + # [(s, i) for i, s in enumerate(SENTENCES)], + # target=self.model.sv.vectors, + # memory=mem, + # ), # ) - # m2._pre_train_calls() - # mem2 = m2._get_thread_working_mem() - - # from fse.models.pooling_inner import train_pooling_cy - - # o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) - - # self.assertEqual(o1, o2) - # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + # self.assertEqual((105, DIM), self.model.sv.vectors.shape) - # def test_hier_pool_cy_w2v_non_negative(self): - # mpool = MaxPooling(W2V_R, hierarchical=True) - # mpool.train(self.sentences) - # self.assertTrue((mpool.sv.vectors >= 0).all()) + # ### Basic Pooling Tests start here - # def test_hier_pool_cy_ft_non_negative(self): - # mpool = MaxPooling(FT_R, hierarchical=True) - # mpool.train(self.sentences) - # self.assertTrue((mpool.sv.vectors >= 0).all()) - - # ### Hierarchical Test + Stride start here - - # def test_hier_pool_stride_train_np_w2v(self): + # def test_pool_train_np_w2v(self): # self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) # mem = self.model._get_thread_working_mem() - # self.model.hierarchical = True - # self.model.window_stride = 5 - # output = train_pooling_np( # self.model, self.sentences, self.model.sv.vectors, mem # ) - # self.model.hierarchical = False - # self.model.window_stride = 1 # self.assertEqual((5, 14), output) - # self.assertTrue((183 == self.model.sv[0]).all()) - # self.assertTrue((231 == self.model.sv[4]).all()) + # self.assertTrue((241 == self.model.sv[0]).all()) + # self.assertTrue((306 == self.model.sv[1]).all()) + # self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) - # def test_hier_pool_stride_train_cy_w2v(self): + # def test_pool_train_cy_w2v(self): # self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) # mem = self.model._get_thread_working_mem() - # self.model.hierarchical = True - # self.model.window_stride = 5 - # from fse.models.pooling_inner import train_pooling_cy # output = train_pooling_cy( # self.model, self.sentences, self.model.sv.vectors, mem # ) - # self.model.hierarchical = False - # self.model.window_stride = 1 # self.assertEqual((5, 14), output) - # self.assertTrue((183 == self.model.sv[0]).all()) - # self.assertTrue((231 == self.model.sv[4]).all()) - - # def test_hier_pool_stride_train_np_ft(self): - # m = MaxPooling(FT, hierarchical=True, window_stride=3) - # m.prep.prepare_vectors( - # sv=m.sv, total_sentences=len(self.sentences), update=False - # ) - # m._pre_train_calls() - # mem = m._get_thread_working_mem() - - # output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) - - # self.assertEqual((5, 19), output) - # self.assertTrue((183 == m.sv[0]).all()) - # self.assertTrue(np.allclose(368871.94, m.sv[2])) - # self.assertTrue(np.allclose(961940.2, m.sv[3])) - - # def test_hier_pool_stride_train_cy_ft(self): - # m = MaxPooling(FT, hierarchical=True, window_stride=3) - # m.prep.prepare_vectors( - # sv=m.sv, total_sentences=len(self.sentences), update=False - # ) - # m._pre_train_calls() - # mem = m._get_thread_working_mem() - - # from fse.models.pooling_inner import train_pooling_cy + # self.assertTrue((241 == self.model.sv[0]).all()) + # self.assertTrue((306 == self.model.sv[1]).all()) + # self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) + + +###### Worked until here + +# def test_pool_train_np_ft(self): +# m = MaxPooling(FT) +# m.prep.prepare_vectors( +# sv=m.sv, total_sentences=len(self.sentences), update=False +# ) +# m._pre_train_calls() +# mem = m._get_thread_working_mem() + +# output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) + +# self.assertEqual((5, 19), output) +# self.assertTrue((241 == m.sv[0]).all()) +# self.assertTrue( +# np.allclose(737413.9, m.sv[2]) +# ) +# self.assertTrue( +# np.allclose(1080970.2, m.sv[3]) +# ) + +# def test_pool_train_cy_ft(self): +# m = MaxPooling(FT) +# m.prep.prepare_vectors( +# sv=m.sv, total_sentences=len(self.sentences), update=False +# ) +# m._pre_train_calls() +# mem = m._get_thread_working_mem() + +# from fse.models.pooling_inner import train_pooling_cy + +# output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) + +# self.assertEqual((5, 19), output) +# self.assertTrue((241 == m.sv[0]).all()) +# self.assertTrue( +# np.allclose(737413.9, m.sv[2]) +# ) +# self.assertTrue( +# np.allclose(1080970.2, m.sv[3]) +# ) + +# def test_pool_cy_equal_np_w2v(self): +# m1 = MaxPooling(W2V) +# m1.prep.prepare_vectors( +# sv=m1.sv, total_sentences=len(self.sentences), update=False +# ) +# m1._pre_train_calls() +# mem1 = m1._get_thread_working_mem() +# o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + +# m2 = MaxPooling(W2V) +# m2.prep.prepare_vectors( +# sv=m2.sv, total_sentences=len(self.sentences), update=False +# ) +# m2._pre_train_calls() +# mem2 = m2._get_thread_working_mem() + +# from fse.models.pooling_inner import train_pooling_cy + +# o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + +# self.assertEqual(o1, o2) +# self.assertTrue((m1.sv.vectors == m2.sv.vectors).all()) + +# def test_pool_cy_equal_np_w2v_random(self): +# w2v = Word2Vec(min_count=1, size=DIM) +# # Random initialization +# w2v.build_vocab(SENTENCES) + +# m1 = MaxPooling(w2v) +# m1.prep.prepare_vectors( +# sv=m1.sv, total_sentences=len(self.sentences), update=False +# ) +# m1._pre_train_calls() +# mem1 = m1._get_thread_working_mem() +# o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + +# m2 = MaxPooling(w2v) +# m2.prep.prepare_vectors( +# sv=m2.sv, total_sentences=len(self.sentences), update=False +# ) +# m2._pre_train_calls() +# mem2 = m2._get_thread_working_mem() + +# from fse.models.pooling_inner import train_pooling_cy + +# o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + +# self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + +# def test_pool_cy_equal_np_ft_random(self): +# ft = FastText(size=20, min_count=1) +# ft.build_vocab(SENTENCES) + +# m1 = MaxPooling(ft) +# m1.prep.prepare_vectors( +# sv=m1.sv, total_sentences=len(self.sentences), update=False +# ) +# m1._pre_train_calls() + +# from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH + +# m1.batch_ngrams = MAX_NGRAMS_IN_BATCH +# mem1 = m1._get_thread_working_mem() +# o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) + +# m2 = MaxPooling(ft) +# m2.prep.prepare_vectors( +# sv=m2.sv, total_sentences=len(self.sentences), update=False +# ) +# m2._pre_train_calls() +# mem2 = m2._get_thread_working_mem() + +# from fse.models.pooling_inner import train_pooling_cy + +# o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) + +# self.assertEqual(o1, o2) +# self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + +# def test_pool_np_w2v_non_negative(self): +# mpool = MaxPooling(W2V_R) +# mpool.train(self.sentences) +# self.assertTrue((mpool.sv.vectors >= 0).all()) + +# def test_pool_np_ft_non_negative(self): +# mpool = MaxPooling(FT_R) +# mpool.train(self.sentences) +# self.assertTrue((mpool.sv.vectors >= 0).all()) + +### Hierarchical Tests start here + +# def test_hier_pool_train_np_w2v(self): +# self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) +# mem = self.model._get_thread_working_mem() + +# self.model.hierarchical = True + +# output = train_pooling_np( +# self.model, self.sentences, self.model.sv.vectors, mem +# ) +# self.model.hierarchical = False + +# self.assertEqual((5, 14), output) +# self.assertTrue((183 == self.model.sv[0]).all()) +# self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) + +# def test_hier_pool_train_cy_w2v(self): +# self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) +# mem = self.model._get_thread_working_mem() + +# self.model.hierarchical = True + +# from fse.models.pooling_inner import train_pooling_cy + +# output = train_pooling_cy( +# self.model, self.sentences, self.model.sv.vectors, mem +# ) +# self.model.hierarchical = False + +# self.assertEqual((5, 14), output) +# self.assertTrue((183 == self.model.sv[0]).all()) +# self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) + +# def test_hier_pool_train_np_ft(self): +# m = MaxPooling(FT, hierarchical=True) +# m.prep.prepare_vectors( +# sv=m.sv, total_sentences=len(self.sentences), update=False +# ) +# m._pre_train_calls() +# mem = m._get_thread_working_mem() + +# output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) + +# self.assertEqual((5, 19), output) +# self.assertTrue((183 == m.sv[0]).all()) +# self.assertTrue(np.allclose(737413.9, m.sv[2])) +# self.assertTrue(np.allclose(1080970.2, m.sv[3])) +# """ +# Note to future self: +# Due to the size of the ngram vectors, +# an ngram at the last position of the senence +# will always be the highest value. +# TODO: This unittest is thus a bit flawed. Maybe fix? +# """ - # output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) +# def test_hier_pool_train_cy_ft(self): +# m = MaxPooling(FT, hierarchical=True) +# m.prep.prepare_vectors( +# sv=m.sv, total_sentences=len(self.sentences), update=False +# ) +# m._pre_train_calls() +# mem = m._get_thread_working_mem() + +# from fse.models.pooling_inner import train_pooling_cy + +# output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) - # self.assertEqual((5, 19), output) - # self.assertTrue((183 == m.sv[0]).all()) - # self.assertTrue(np.allclose(368871.94, m.sv[2])) - # self.assertTrue(np.allclose(961940.2, m.sv[3])) +# self.assertEqual((5, 19), output) +# self.assertTrue((183 == m.sv[0]).all()) +# self.assertTrue(np.allclose(737413.9, m.sv[2])) +# self.assertTrue(np.allclose(1080970.2, m.sv[3])) + +# def test_hier_pool_cy_equal_np_w2v_random(self): +# w2v = Word2Vec(min_count=1, size=DIM) +# # Random initialization +# w2v.build_vocab(SENTENCES) + +# m1 = MaxPooling(w2v, hierarchical=True) +# m1.prep.prepare_vectors( +# sv=m1.sv, total_sentences=len(self.sentences), update=False +# ) +# m1._pre_train_calls() +# mem1 = m1._get_thread_working_mem() +# o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + +# m2 = MaxPooling(w2v, hierarchical=True) +# m2.prep.prepare_vectors( +# sv=m2.sv, total_sentences=len(self.sentences), update=False +# ) +# m2._pre_train_calls() +# mem2 = m2._get_thread_working_mem() + +# from fse.models.pooling_inner import train_pooling_cy + +# o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + +# self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + +# def test_hier_pool_cy_equal_np_ft_random(self): +# ft = FastText(size=20, min_count=1) +# ft.build_vocab(SENTENCES) + +# m1 = MaxPooling(ft, hierarchical=True) +# m1.prep.prepare_vectors( +# sv=m1.sv, total_sentences=len(self.sentences), update=False +# ) +# m1._pre_train_calls() + +# from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH + +# m1.batch_ngrams = MAX_NGRAMS_IN_BATCH +# mem1 = m1._get_thread_working_mem() +# o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) + +# m2 = MaxPooling(ft, hierarchical=True) +# m2.prep.prepare_vectors( +# sv=m2.sv, total_sentences=len(self.sentences), update=False +# ) +# m2._pre_train_calls() +# mem2 = m2._get_thread_working_mem() + +# from fse.models.pooling_inner import train_pooling_cy + +# o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) + +# self.assertEqual(o1, o2) +# self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + +# def test_hier_pool_cy_w2v_non_negative(self): +# mpool = MaxPooling(W2V_R, hierarchical=True) +# mpool.train(self.sentences) +# self.assertTrue((mpool.sv.vectors >= 0).all()) + +# def test_hier_pool_cy_ft_non_negative(self): +# mpool = MaxPooling(FT_R, hierarchical=True) +# mpool.train(self.sentences) +# self.assertTrue((mpool.sv.vectors >= 0).all()) + +# ### Hierarchical Test + Stride start here + +# def test_hier_pool_stride_train_np_w2v(self): +# self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) +# mem = self.model._get_thread_working_mem() + +# self.model.hierarchical = True +# self.model.window_stride = 5 + +# output = train_pooling_np( +# self.model, self.sentences, self.model.sv.vectors, mem +# ) +# self.model.hierarchical = False +# self.model.window_stride = 1 + +# self.assertEqual((5, 14), output) +# self.assertTrue((183 == self.model.sv[0]).all()) +# self.assertTrue((231 == self.model.sv[4]).all()) + +# def test_hier_pool_stride_train_cy_w2v(self): +# self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) +# mem = self.model._get_thread_working_mem() - # def test_hier_pool_stride_cy_equal_np_w2v_random(self): - # w2v = Word2Vec(min_count=1, size=DIM) - # # Random initialization - # w2v.build_vocab(SENTENCES) +# self.model.hierarchical = True +# self.model.window_stride = 5 - # m1 = MaxPooling(w2v, hierarchical=True, window_stride=4) - # m1.prep.prepare_vectors( - # sv=m1.sv, total_sentences=len(self.sentences), update=False - # ) - # m1._pre_train_calls() - # mem1 = m1._get_thread_working_mem() - # o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) - - # m2 = MaxPooling(w2v, hierarchical=True, window_stride=4) - # m2.prep.prepare_vectors( - # sv=m2.sv, total_sentences=len(self.sentences), update=False - # ) - # m2._pre_train_calls() - # mem2 = m2._get_thread_working_mem() - - # from fse.models.pooling_inner import train_pooling_cy - - # o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) - - # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - - # def test_hier_pool_stride_cy_equal_np_ft_random(self): - # ft = FastText(size=20, min_count=1) - # ft.build_vocab(SENTENCES) - - # m1 = MaxPooling(ft, hierarchical=True, window_stride=5) - # m1.prep.prepare_vectors( - # sv=m1.sv, total_sentences=len(self.sentences), update=False - # ) - # m1._pre_train_calls() - - # from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH - - # m1.batch_ngrams = MAX_NGRAMS_IN_BATCH - # mem1 = m1._get_thread_working_mem() - # o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - - # m2 = MaxPooling(ft, hierarchical=True, window_stride=5) - # m2.prep.prepare_vectors( - # sv=m2.sv, total_sentences=len(self.sentences), update=False - # ) - # m2._pre_train_calls() - # mem2 = m2._get_thread_working_mem() - - # from fse.models.pooling_inner import train_pooling_cy +# from fse.models.pooling_inner import train_pooling_cy + +# output = train_pooling_cy( +# self.model, self.sentences, self.model.sv.vectors, mem +# ) +# self.model.hierarchical = False +# self.model.window_stride = 1 + +# self.assertEqual((5, 14), output) +# self.assertTrue((183 == self.model.sv[0]).all()) +# self.assertTrue((231 == self.model.sv[4]).all()) + +# def test_hier_pool_stride_train_np_ft(self): +# m = MaxPooling(FT, hierarchical=True, window_stride=3) +# m.prep.prepare_vectors( +# sv=m.sv, total_sentences=len(self.sentences), update=False +# ) +# m._pre_train_calls() +# mem = m._get_thread_working_mem() - # o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) +# output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) + +# self.assertEqual((5, 19), output) +# self.assertTrue((183 == m.sv[0]).all()) +# self.assertTrue(np.allclose(368871.94, m.sv[2])) +# self.assertTrue(np.allclose(961940.2, m.sv[3])) - # self.assertEqual(o1, o2) - # self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) +# def test_hier_pool_stride_train_cy_ft(self): +# m = MaxPooling(FT, hierarchical=True, window_stride=3) +# m.prep.prepare_vectors( +# sv=m.sv, total_sentences=len(self.sentences), update=False +# ) +# m._pre_train_calls() +# mem = m._get_thread_working_mem() + +# from fse.models.pooling_inner import train_pooling_cy + +# output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) + +# self.assertEqual((5, 19), output) +# self.assertTrue((183 == m.sv[0]).all()) +# self.assertTrue(np.allclose(368871.94, m.sv[2])) +# self.assertTrue(np.allclose(961940.2, m.sv[3])) - # def test_hier_pool_stride_np_w2v_non_negative(self): - # mpool = MaxPooling(W2V_R, hierarchical=True, window_stride=4) - # mpool.train(self.sentences) - # self.assertTrue((mpool.sv.vectors >= 0).all()) +# def test_hier_pool_stride_cy_equal_np_w2v_random(self): +# w2v = Word2Vec(min_count=1, size=DIM) +# # Random initialization +# w2v.build_vocab(SENTENCES) + +# m1 = MaxPooling(w2v, hierarchical=True, window_stride=4) +# m1.prep.prepare_vectors( +# sv=m1.sv, total_sentences=len(self.sentences), update=False +# ) +# m1._pre_train_calls() +# mem1 = m1._get_thread_working_mem() +# o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) - # def test_hier_pool_stride_np_ft_non_negative(self): - # mpool = MaxPooling(FT_R, hierarchical=True, window_stride=4) - # mpool.train(self.sentences) - # self.assertTrue((mpool.sv.vectors >= 0).all()) +# m2 = MaxPooling(w2v, hierarchical=True, window_stride=4) +# m2.prep.prepare_vectors( +# sv=m2.sv, total_sentences=len(self.sentences), update=False +# ) +# m2._pre_train_calls() +# mem2 = m2._get_thread_working_mem() + +# from fse.models.pooling_inner import train_pooling_cy + +# o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + +# self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + +# def test_hier_pool_stride_cy_equal_np_ft_random(self): +# ft = FastText(size=20, min_count=1) +# ft.build_vocab(SENTENCES) + +# m1 = MaxPooling(ft, hierarchical=True, window_stride=5) +# m1.prep.prepare_vectors( +# sv=m1.sv, total_sentences=len(self.sentences), update=False +# ) +# m1._pre_train_calls() + +# from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH + +# m1.batch_ngrams = MAX_NGRAMS_IN_BATCH +# mem1 = m1._get_thread_working_mem() +# o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) + +# m2 = MaxPooling(ft, hierarchical=True, window_stride=5) +# m2.prep.prepare_vectors( +# sv=m2.sv, total_sentences=len(self.sentences), update=False +# ) +# m2._pre_train_calls() +# mem2 = m2._get_thread_working_mem() + +# from fse.models.pooling_inner import train_pooling_cy + +# o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) + +# self.assertEqual(o1, o2) +# self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + +# def test_hier_pool_stride_np_w2v_non_negative(self): +# mpool = MaxPooling(W2V_R, hierarchical=True, window_stride=4) +# mpool.train(self.sentences) +# self.assertTrue((mpool.sv.vectors >= 0).all()) + +# def test_hier_pool_stride_np_ft_non_negative(self): +# mpool = MaxPooling(FT_R, hierarchical=True, window_stride=4) +# mpool.train(self.sentences) +# self.assertTrue((mpool.sv.vectors >= 0).all()) if __name__ == "__main__": diff --git a/fse/test/test_utils.py b/fse/test/test_utils.py index 4ef3c62..40ca863 100644 --- a/fse/test/test_utils.py +++ b/fse/test/test_utils.py @@ -7,13 +7,13 @@ from shared_imports import * from fse.models.utils import ( - compute_principal_components, + compute_principal_components, remove_principal_components, get_ft_word_vector, ) + class TestUtils(unittest.TestCase): - def test_compute_components(self): m = np.random.uniform(size=(500, 10)).astype(np.float32) out = compute_principal_components(vectors=m) @@ -74,6 +74,7 @@ def test_madvise(self): def test_ft_word_vector(self): from fse.models.average import Average + model = Average(FT_DET) word = "123456789" @@ -82,6 +83,7 @@ def test_ft_word_vector(self): self.assertIsInstance(vec, np.ndarray) self.assertTrue(np.allclose(max(vec), 979980.94)) + if __name__ == "__main__": logging.basicConfig( format="%(asctime)s : %(levelname)s : %(message)s", level=logging.DEBUG From ccebd2808c97755d558a9901bff614ac5eb44eec Mon Sep 17 00:00:00 2001 From: OliverB Date: Fri, 22 May 2020 14:29:07 +0200 Subject: [PATCH 46/63] Fixed shared imports --- ...red_imports.py => model_shared_imports.py} | 5 +++ fse/test/test_average.py | 3 +- fse/test/test_pooling.py | 41 +------------------ fse/test/test_sif.py | 30 +++----------- fse/test/test_usif.py | 27 +++--------- fse/test/test_utils.py | 2 +- 6 files changed, 20 insertions(+), 88 deletions(-) rename fse/test/{shared_imports.py => model_shared_imports.py} (87%) diff --git a/fse/test/shared_imports.py b/fse/test/model_shared_imports.py similarity index 87% rename from fse/test/shared_imports.py rename to fse/test/model_shared_imports.py index 942f402..4202dcb 100644 --- a/fse/test/shared_imports.py +++ b/fse/test/model_shared_imports.py @@ -8,6 +8,7 @@ from gensim.models import Word2Vec, FastText from fse.models.base_s2v import EPS +from fse.inputs import IndexedLineDocument logger = logging.getLogger(__name__) @@ -28,6 +29,10 @@ W2V_RNG = Word2Vec(min_count=1, size=DIM) W2V_RNG.build_vocab(SENTENCES) +# Random W2V Large +W2V_RNG_LRG = Word2Vec(min_count=1, size=50) +W2V_RNG_LRG.build_vocab(SENTENCES) + # Deterministic FT FT_DET = FastText(min_count=1, size=DIM) FT_DET.build_vocab(SENTENCES) diff --git a/fse/test/test_average.py b/fse/test/test_average.py index 10c719a..bc1dbfe 100644 --- a/fse/test/test_average.py +++ b/fse/test/test_average.py @@ -9,11 +9,10 @@ Automated tests for checking the average model. """ -from shared_imports import * +from fse.test.model_shared_imports import * from fse.models.average import Average, train_average_np - class TestAverageFunctions(unittest.TestCase): def setUp(self): self.sentences = [ diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index 050fbeb..b5cad88 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -5,46 +5,9 @@ # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. -""" -Automated tests for checking the model. -""" - -import logging -import unittest - -from pathlib import Path - -import numpy as np - from fse.models.pooling import MaxPooling, train_pooling_np -from fse.models.base_s2v import EPS - -from gensim.models import Word2Vec, FastText - -logger = logging.getLogger(__name__) - -CORPUS = Path("fse/test/test_data/test_sentences.txt") -DIM = 5 -W2V = Word2Vec(min_count=1, size=DIM) -with open(CORPUS, "r") as f: - SENTENCES = [l.split() for i, l in enumerate(f)] -W2V.build_vocab(SENTENCES) -W2V.wv.vectors[:,] = np.arange(len(W2V.wv.vectors), dtype=np.float32)[:, None] - -W2V_R = Word2Vec(min_count=1, size=DIM) -W2V_R.build_vocab(SENTENCES) - -FT = FastText(min_count=1, size=DIM) -FT.build_vocab(SENTENCES) -FT.wv.vectors[:,] = np.arange(len(FT.wv.vectors), dtype=np.float32)[:, None] -FT.wv.vectors_vocab = FT.wv.vectors -FT.wv.vectors_ngrams[:,] = np.arange(len(FT.wv.vectors_ngrams), dtype=np.float32)[ - :, None -] - -FT_R = FastText(min_count=1, size=DIM) -FT_R.build_vocab(SENTENCES) +from fse.test.model_shared_imports import * class TestPoolingFunctions(unittest.TestCase): def setUp(self): @@ -56,7 +19,7 @@ def setUp(self): "this is a longer test sentence test longer sentences".split(), ] self.sentences = [(s, i) for i, s in enumerate(self.sentences)] - self.model = MaxPooling(W2V) + self.model = MaxPooling(W2V_DET) self.model.prep.prepare_vectors( sv=self.model.sv, total_sentences=len(self.sentences), update=False ) diff --git a/fse/test/test_sif.py b/fse/test/test_sif.py index e44a704..a6068cd 100644 --- a/fse/test/test_sif.py +++ b/fse/test/test_sif.py @@ -4,46 +4,28 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -import logging -import unittest - -from pathlib import Path - -import numpy as np - from fse.models.sif import ( SIF, compute_principal_components, remove_principal_components, ) -from fse.inputs import IndexedLineDocument - -from gensim.models import Word2Vec - -logger = logging.getLogger(__name__) - -CORPUS = Path("fse/test/test_data/test_sentences.txt") -DIM = 50 -W2V = Word2Vec(min_count=1, size=DIM) -with open(CORPUS, "r") as f: - SENTENCES = [l.split() for i, l in enumerate(f)] -W2V.build_vocab(SENTENCES) +from fse.test.model_shared_imports import * class TestSIFFunctions(unittest.TestCase): def setUp(self): self.sentences = IndexedLineDocument(CORPUS) - self.model = SIF(W2V, lang_freq="en") + self.model = SIF(W2V_RNG_LRG, lang_freq="en") def test_parameter_sanity(self): with self.assertRaises(ValueError): - m = SIF(W2V, alpha=-1) + m = SIF(W2V_RNG_LRG, alpha=-1) m._check_parameter_sanity() with self.assertRaises(ValueError): - m = SIF(W2V, components=-1) + m = SIF(W2V_RNG_LRG, components=-1) m._check_parameter_sanity() with self.assertRaises(ValueError): - m = SIF(W2V) + m = SIF(W2V_RNG_LRG) m.word_weights = np.ones_like(m.word_weights) + 2 m._check_parameter_sanity() @@ -121,7 +103,7 @@ def test_train(self): self.assertEqual(2, len(self.model.svd_res)) def test_save_issue(self): - model = SIF(W2V) + model = SIF(W2V_RNG_LRG) model.train(self.sentences) p = Path("fse/test/test_data/test_emb.model") diff --git a/fse/test/test_usif.py b/fse/test/test_usif.py index b9ba90d..43c8a14 100644 --- a/fse/test/test_usif.py +++ b/fse/test/test_usif.py @@ -4,39 +4,22 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -import logging -import unittest - -from pathlib import Path - -import numpy as np - from fse.models.usif import uSIF -from fse.inputs import IndexedLineDocument - -from gensim.models import Word2Vec - -logger = logging.getLogger(__name__) -CORPUS = Path("fse/test/test_data/test_sentences.txt") -DIM = 50 -W2V = Word2Vec(min_count=1, size=DIM) -with open(CORPUS, "r") as f: - SENTENCES = [l.split() for i, l in enumerate(f)] -W2V.build_vocab(SENTENCES) +from fse.test.model_shared_imports import * class TestuSIFFunctions(unittest.TestCase): def setUp(self): self.sentences = IndexedLineDocument(CORPUS) - self.model = uSIF(W2V, lang_freq="en") + self.model = uSIF(W2V_RNG_LRG, lang_freq="en") def test_parameter_sanity(self): with self.assertRaises(ValueError): - m = uSIF(W2V, length=0) + m = uSIF(W2V_RNG_LRG, length=0) m._check_parameter_sanity() with self.assertRaises(ValueError): - m = uSIF(W2V, components=-1, length=11) + m = uSIF(W2V_RNG_LRG, components=-1, length=11) m._check_parameter_sanity() def test_pre_train_calls(self): @@ -113,7 +96,7 @@ def test_train(self): self.assertTrue(np.isfinite(self.model.sv.vectors).all()) def test_broken_vocab(self): - w2v = Word2Vec(min_count=1, size=DIM) + w2v = Word2Vec(min_count=1, size=50) w2v.build_vocab([l.split() for l in open(CORPUS, "r")]) for k in w2v.wv.vocab: w2v.wv.vocab[k].count = np.nan diff --git a/fse/test/test_utils.py b/fse/test/test_utils.py index 40ca863..c882e33 100644 --- a/fse/test/test_utils.py +++ b/fse/test/test_utils.py @@ -4,7 +4,7 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -from shared_imports import * +from fse.test.model_shared_imports import * from fse.models.utils import ( compute_principal_components, From d74a1386d3b16bc298e692c2e356df4c9e6ce2b1 Mon Sep 17 00:00:00 2001 From: OliverB Date: Tue, 4 Aug 2020 16:06:22 +0200 Subject: [PATCH 47/63] fixed shared imports --- fse/test/model_shared_imports.py | 1 + 1 file changed, 1 insertion(+) diff --git a/fse/test/model_shared_imports.py b/fse/test/model_shared_imports.py index 4202dcb..e76dc4f 100644 --- a/fse/test/model_shared_imports.py +++ b/fse/test/model_shared_imports.py @@ -16,6 +16,7 @@ CORPUS = Path("fse/test/test_data/test_sentences.txt") with open(CORPUS, "r") as f: SENTENCES = [l.split() for i, l in enumerate(f)] +ENUM_SENTENCES = [(l, i) for i, l in enumerate(SENTENCES)] # Models DIM = 5 From d8633ab0152addb28f07ba630f99499c91c113c3 Mon Sep 17 00:00:00 2001 From: OliverB Date: Tue, 4 Aug 2020 16:08:09 +0200 Subject: [PATCH 48/63] added todo --- fse/models/sentencevectors.py | 1 + 1 file changed, 1 insertion(+) diff --git a/fse/models/sentencevectors.py b/fse/models/sentencevectors.py index 7590708..9036ecc 100644 --- a/fse/models/sentencevectors.py +++ b/fse/models/sentencevectors.py @@ -74,6 +74,7 @@ def __init__(self, vector_size: int, mapfile_path: str = None): [ ] Lookup-Functionality (via defaultdict) [ ] Get vector: Not really memory friendly [ ] Show which words are in vocabulary + [ ] Asses empty vectors (via EPS sum) """ def __getitem__(self, entities: int) -> ndarray: From 77ab0985f200f57f41e683440346091a6ca9db14 Mon Sep 17 00:00:00 2001 From: OliverB Date: Tue, 4 Aug 2020 16:19:27 +0200 Subject: [PATCH 49/63] updated readme --- README.md | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/README.md b/README.md index 50e1912..870019f 100644 --- a/README.md +++ b/README.md @@ -19,8 +19,8 @@ Features Find the corresponding blog post(s) here: -- [Visualizing 100,000 Amazon Products](https://towardsdatascience.com/vis-amz-83dea6fcb059) -- [Sentence Embeddings. Fast, please!](https://towardsdatascience.com/fse-2b1ffa791cf9) +- [Visualizing 100,000 Amazon Products](https://towardsdatascience.com/vis-amz-83dea6fcb059) (Note: The code may be outdated) +- [Sentence Embeddings. Fast, please!](https://towardsdatascience.com/fse-2b1ffa791cf9) (Note: The code may be outdated) **fse** implements multiple algorithms for sentence embeddings. You can choose between *unweighted sentence averages*, *smooth inverse frequency averages*, *unsupervised smooth inverse frequency averages*, and *max pooling*. All models support hierarchical estimation, similar to convolutional filters in CNNs. From 158dd4875f50b966601a08b2523226ccd81f17b4 Mon Sep 17 00:00:00 2001 From: OliverB Date: Tue, 4 Aug 2020 16:26:27 +0200 Subject: [PATCH 50/63] Changed error message --- fse/models/sif.py | 2 +- fse/models/usif.py | 2 +- 2 files changed, 2 insertions(+), 2 deletions(-) diff --git a/fse/models/sif.py b/fse/models/sif.py index c79eb8b..55bb95f 100644 --- a/fse/models/sif.py +++ b/fse/models/sif.py @@ -151,5 +151,5 @@ def _compute_sif_weights(self): raise RuntimeError( "Encountered nan values. " "This likely happens because the word frequency information is wrong/missing. " - "Consider restarting using lang_freq argument to infer frequency. " + "Restart using lang_freq argument to infer frequency." ) diff --git a/fse/models/usif.py b/fse/models/usif.py index df1a74b..59c29ba 100644 --- a/fse/models/usif.py +++ b/fse/models/usif.py @@ -172,5 +172,5 @@ def _compute_usif_weights(self): raise RuntimeError( "Encountered nan values. " "This likely happens because the word frequency information is wrong/missing. " - "Consider restarting using lang_freq argument to infer frequency. " + "Restart using lang_freq argument to infer frequency." ) From b36306dcde8775416d03685cf15339f81780873e Mon Sep 17 00:00:00 2001 From: OliverB Date: Tue, 4 Aug 2020 18:08:27 +0200 Subject: [PATCH 51/63] Updated todo --- fse/models/base_s2v.py | 85 +++++++++++++++++++++++------------ fse/models/sentencevectors.py | 26 ----------- 2 files changed, 57 insertions(+), 54 deletions(-) diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index 2ffadac..6cb477e 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -77,7 +77,6 @@ EPS = finfo(REAL).eps - class BaseSentence2VecModel(SaveLoad): def __init__( self, @@ -127,47 +126,77 @@ def __init__( """ TODO: - [ ] global: - [ ] windows support - [ ] documentation + [ ] Global: + Immediate: + [ ] Documentation + [ ] Mail address update + [ ] remove wv_mapfile_path? + [ ] modifiable sv_mapfile_path? + [ ] Python 3.8 Travis Build + [ ] Possible to make a slim-save version and discard very large files? + [ ] Check with latest Gensim version! + Postponed: [ ] more benchmarks [ ] more speed benchmarks -> worker support still somewhat weird - [-] remove wv_mapfile_path? - [ ] modifiable sv_mapfile_path? - - [ ] models: - [ ] Check feasibility first + [ ] windows support - [ ] Average: + [ ] Models: + Immediate + [ ] pre_inference_calls + [ ] base_iterator: [ ] Hierarchical Average [ ] Stride Parameter: https://adeshpande3.github.io/A-Beginner%27s-Guide-To-Understanding-Convolutional-Neural-Networks-Part-2/ - [X] Max Pooling - [X] Hierarchical Max Pooling - [X] Stride Paramter: https://adeshpande3.github.io/A-Beginner%27s-Guide-To-Understanding-Convolutional-Neural-Networks-Part-2/ - + [ ] Average: + [ ] inherit from base_iterator + [ ] avg core routines in cyhton + [ ] Max Pooling + [ ] inherit from base_iterator + [ ] avg core routines in cyhton + [ ] Check for further simple models + [ ] Min Pooling + Lower: + [ ] GEM: https://arxiv.org/pdf/1810.00438.pdf + [ ] ALaCarte + + Postponed: [ ] discrete cosine transform -> somewhat easy, questionable [ ] valve -> unclear, not cited enough [ ] power-means embedding -> very large dimensionalty [ ] z-score transformation is quite nice - - [ ] sentencevectors: - [ ] remove the L1 normalization or make it optional? - [X] similar_by_sentence model type check - [ ] approximate NN search for large files - [X] compare ANN libraries - [X] ease-of-use - [X] dependencies - [X] compatibility - [X] memory-usage - --> Annoy - - [ ] unittests: + + [ ] Sentencevectors: + Global: + [ ] Remove normalized vector files and replace with NN + ANN: --> (Annoy, with Option for Google ScANN?) + [ ] Only construct index when when calling most_similar method + [ ] Logging of index speed + [ ] Save and load of index + [ ] Assert that index and vectors are of equal size + [ ] Paramters must be tunable afterwards + [ ] Method to reconstruct index + [ ] How does the index saving comply with SaveLoad? + [ ] Write unittests? + Brute: + [ ] Keep access to default method + [ ] Make ANN Search the default?! --> Results? + [ ] Throw warning for large datasets for vector norm init + [ ] Maybe throw warning if exceeds RAM size of the embedding + normalization + Other: + [ ] L2 Distance + [ ] L1 Distance + [ ] Correlation (Power Score Correlation?) + [ ] Lookup-Functionality (via defaultdict) + [ ] Get vector: Not really memory friendly + [ ] Show which words are in vocabulary + [ ] Asses empty vectors (via EPS sum) + [ ] Z-Score Transformation from Power-Means Embedding? --> Benefit? + + [ ] Tests: [ ] Check that custom index is numerical only [ ] Raise [ ] Unittest """ - set_madvise_for_mmap() self.workers = int(workers) diff --git a/fse/models/sentencevectors.py b/fse/models/sentencevectors.py index 9036ecc..c95c7b1 100644 --- a/fse/models/sentencevectors.py +++ b/fse/models/sentencevectors.py @@ -51,32 +51,6 @@ def __init__(self, vector_size: int, mapfile_path: str = None): self.mapfile_path = Path(mapfile_path) if mapfile_path is not None else None self.mapfile_shape = None - """ - Thoughts on additinal features: - ANN: - [ ] Only construct index when when calling most_similar method - [ ] Logging of index speed - [ ] Save and load of index - [ ] Assert that index and vectors are of equal size - [ ] Paramters must be tunable afterwards - [ ] Method to reconstruct index - [ ] How does the index saving comply with SaveLoad? - [ ] Write unittests? - Brute: - [ ] Keep access to default method - [ ] Make ANN Search the default?! --> Results? - [ ] Throw warning for large datasets for vector norm init - [ ] Maybe throw warning if exceeds RAM size of the embedding + normalization - Other: - [ ] L2 Distance - [ ] L1 Distance - [ ] Correlation - [ ] Lookup-Functionality (via defaultdict) - [ ] Get vector: Not really memory friendly - [ ] Show which words are in vocabulary - [ ] Asses empty vectors (via EPS sum) - """ - def __getitem__(self, entities: int) -> ndarray: """Get vector representation of `entities`. From 1d01d1d48d0aafb7e04e2801f8351bccaa06f585 Mon Sep 17 00:00:00 2001 From: OliverB Date: Tue, 4 Aug 2020 18:08:36 +0200 Subject: [PATCH 52/63] Updated todo average --- fse/test/test_average.py | 23 ++++++++++++----------- 1 file changed, 12 insertions(+), 11 deletions(-) diff --git a/fse/test/test_average.py b/fse/test/test_average.py index bc1dbfe..26aa39d 100644 --- a/fse/test/test_average.py +++ b/fse/test/test_average.py @@ -240,17 +240,18 @@ def test_check_parameter_sanity(self): with self.assertRaises(ValueError): se._check_parameter_sanity() - se = Average(W2V_DET, window_size=0) - with self.assertRaises(ValueError): - se._check_parameter_sanity() - - se = Average(W2V_DET, window_size=3, window_stride=0) - with self.assertRaises(ValueError): - se._check_parameter_sanity() - - se = Average(W2V_DET, window_size=3, window_stride=4) - with self.assertRaises(ValueError): - se._check_parameter_sanity() + # TODO: + # se = Average(W2V_DET, window_size=0) + # with self.assertRaises(ValueError): + # se._check_parameter_sanity() + + # se = Average(W2V_DET, window_size=3, window_stride=0) + # with self.assertRaises(ValueError): + # se._check_parameter_sanity() + + # se = Average(W2V_DET, window_size=3, window_stride=4) + # with self.assertRaises(ValueError): + # se._check_parameter_sanity() if __name__ == "__main__": From f4f3fe533e5c7b0e0624f74c09163943cb6b3cab Mon Sep 17 00:00:00 2001 From: OliverB Date: Tue, 4 Aug 2020 18:19:10 +0200 Subject: [PATCH 53/63] Readme + todo --- README.md | 12 ++++++------ fse/models/base_s2v.py | 3 ++- 2 files changed, 8 insertions(+), 7 deletions(-) diff --git a/README.md b/README.md index 870019f..dbc8cc8 100644 --- a/README.md +++ b/README.md @@ -4,6 +4,7 @@ Downloads Language grade: Python Code style: black +License: GPL3

Fast Sentence Embeddings (fse) @@ -17,11 +18,6 @@ If you want to support fse, take a quick [survey](https://forms.gle/8uSU323fWUVt Features ------------ -Find the corresponding blog post(s) here: - -- [Visualizing 100,000 Amazon Products](https://towardsdatascience.com/vis-amz-83dea6fcb059) (Note: The code may be outdated) -- [Sentence Embeddings. Fast, please!](https://towardsdatascience.com/fse-2b1ffa791cf9) (Note: The code may be outdated) - **fse** implements multiple algorithms for sentence embeddings. You can choose between *unweighted sentence averages*, *smooth inverse frequency averages*, *unsupervised smooth inverse frequency averages*, and *max pooling*. All models support hierarchical estimation, similar to convolutional filters in CNNs. @@ -61,6 +57,10 @@ Key features of **fse** are: I regularly observe 300k-500k sentences/s for preprocessed data on my Macbook (2016). Visit **Tutorial.ipynb** for an example. +Find the corresponding blog post(s) here: + +- [Visualizing 100,000 Amazon Products](https://towardsdatascience.com/vis-amz-83dea6fcb059) (Note: The code may be outdated) +- [Sentence Embeddings. Fast, please!](https://towardsdatascience.com/fse-2b1ffa791cf9) (Note: The code may be outdated) Installation ------------ @@ -104,7 +104,7 @@ for a corpus. model.sv.similarity(0,1) fse offers multi-thread support out of the box. However, for most -applications a *single thread will most likely be sufficient*. +applications a **single** *thread will most likely be sufficient*. Within the folder nootebooks you can find the following guides: diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index 6cb477e..b9c9e24 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -125,7 +125,6 @@ def __init__( """ TODO: - [ ] Global: Immediate: [ ] Documentation @@ -135,6 +134,8 @@ def __init__( [ ] Python 3.8 Travis Build [ ] Possible to make a slim-save version and discard very large files? [ ] Check with latest Gensim version! + [ ] Add Contribution guideline + [ ] Updated performance table and vary pooling size Postponed: [ ] more benchmarks [ ] more speed benchmarks -> worker support still somewhat weird From 4660937c7a46eeaafd556138bdb9cb69ea200848 Mon Sep 17 00:00:00 2001 From: OliverB Date: Wed, 5 Aug 2020 10:27:22 +0200 Subject: [PATCH 54/63] Updated mail --- README.md | 2 +- fse/inputs.py | 2 +- fse/models/average.py | 2 +- fse/models/average_inner.pxd | 2 +- fse/models/average_inner.pyx | 2 +- fse/models/base_s2v.py | 5 +++-- fse/models/pooling.py | 2 +- fse/models/pooling_inner.pyx | 2 +- fse/models/sentencevectors.py | 2 +- fse/models/sif.py | 2 +- fse/models/usif.py | 2 +- fse/models/utils.py | 2 +- fse/test/test_average.py | 2 +- fse/test/test_base_s2v.py | 2 +- fse/test/test_inputs.py | 2 +- fse/test/test_pooling.py | 2 +- fse/test/test_sentencevectors.py | 2 +- fse/test/test_sif.py | 2 +- fse/test/test_usif.py | 2 +- fse/test/test_utils.py | 2 +- setup.py | 2 +- 21 files changed, 23 insertions(+), 22 deletions(-) diff --git a/README.md b/README.md index dbc8cc8..812b422 100644 --- a/README.md +++ b/README.md @@ -198,7 +198,7 @@ Proceedings of the 3rd Workshop on Representation Learning for NLP. (Toulon, Fra Copyright ------------- -Author: Oliver Borchers +Author: Oliver Borchers Copyright (C) 2020 Oliver Borchers diff --git a/fse/inputs.py b/fse/inputs.py index 690c57d..eb52ef5 100644 --- a/fse/inputs.py +++ b/fse/inputs.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers from typing import MutableSequence diff --git a/fse/models/average.py b/fse/models/average.py index 21640a4..52dc8f7 100644 --- a/fse/models/average.py +++ b/fse/models/average.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/models/average_inner.pxd b/fse/models/average_inner.pxd index 2a1ac18..f5e984a 100644 --- a/fse/models/average_inner.pxd +++ b/fse/models/average_inner.pxd @@ -4,7 +4,7 @@ # cython: embedsignature=True # coding: utf-8 -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/models/average_inner.pyx b/fse/models/average_inner.pyx index bc85f28..09820e5 100644 --- a/fse/models/average_inner.pyx +++ b/fse/models/average_inner.pyx @@ -5,7 +5,7 @@ # cython: embedsignature=True # coding: utf-8 -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index b9c9e24..6e49c4d 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. @@ -128,7 +128,7 @@ def __init__( [ ] Global: Immediate: [ ] Documentation - [ ] Mail address update + [X] Mail address update [ ] remove wv_mapfile_path? [ ] modifiable sv_mapfile_path? [ ] Python 3.8 Travis Build @@ -145,6 +145,7 @@ def __init__( Immediate [ ] pre_inference_calls [ ] base_iterator: + [ ] Rewrite VecsConfig to combine FT & W2V [ ] Hierarchical Average [ ] Stride Parameter: https://adeshpande3.github.io/A-Beginner%27s-Guide-To-Understanding-Convolutional-Neural-Networks-Part-2/ [ ] Average: diff --git a/fse/models/pooling.py b/fse/models/pooling.py index c1e6a74..b98670a 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/models/pooling_inner.pyx b/fse/models/pooling_inner.pyx index f01bb66..54e9c0a 100644 --- a/fse/models/pooling_inner.pyx +++ b/fse/models/pooling_inner.pyx @@ -5,7 +5,7 @@ # cython: embedsignature=True # coding: utf-8 -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/models/sentencevectors.py b/fse/models/sentencevectors.py index c95c7b1..e79b1ee 100644 --- a/fse/models/sentencevectors.py +++ b/fse/models/sentencevectors.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/models/sif.py b/fse/models/sif.py index 55bb95f..6b727df 100644 --- a/fse/models/sif.py +++ b/fse/models/sif.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/models/usif.py b/fse/models/usif.py index 59c29ba..a7b7449 100644 --- a/fse/models/usif.py +++ b/fse/models/usif.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/models/utils.py b/fse/models/utils.py index da0dc09..1da633b 100644 --- a/fse/models/utils.py +++ b/fse/models/utils.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/test/test_average.py b/fse/test/test_average.py index 26aa39d..862daf0 100644 --- a/fse/test/test_average.py +++ b/fse/test/test_average.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/test/test_base_s2v.py b/fse/test/test_base_s2v.py index 2915ea9..7df4043 100644 --- a/fse/test/test_base_s2v.py +++ b/fse/test/test_base_s2v.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/test/test_inputs.py b/fse/test/test_inputs.py index ee6b4de..fb96809 100644 --- a/fse/test/test_inputs.py +++ b/fse/test/test_inputs.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index b5cad88..1ec9cb0 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/test/test_sentencevectors.py b/fse/test/test_sentencevectors.py index b740558..fec416c 100644 --- a/fse/test/test_sentencevectors.py +++ b/fse/test/test_sentencevectors.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. diff --git a/fse/test/test_sif.py b/fse/test/test_sif.py index a6068cd..fa772ef 100644 --- a/fse/test/test_sif.py +++ b/fse/test/test_sif.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers from fse.models.sif import ( diff --git a/fse/test/test_usif.py b/fse/test/test_usif.py index 43c8a14..9ca8e42 100644 --- a/fse/test/test_usif.py +++ b/fse/test/test_usif.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers from fse.models.usif import uSIF diff --git a/fse/test/test_utils.py b/fse/test/test_utils.py index c882e33..0396c5b 100644 --- a/fse/test/test_utils.py +++ b/fse/test/test_utils.py @@ -1,7 +1,7 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers +# Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers from fse.test.model_shared_imports import * diff --git a/setup.py b/setup.py index 2a121c2..35f6d85 100755 --- a/setup.py +++ b/setup.py @@ -93,7 +93,7 @@ def finalize_options(self): version="0.1.16", description="Fast Sentence Embeddings for Gensim", author=u"Oliver Borchers", - author_email="borchers@bwl.uni-mannheim.de", + author_email="oliver-borchers@outlook.de", url="https://github.com/oborchers/Fast_Sentence_Embeddings", license="GPL-3.0", ext_modules=[ From 885805ce4da3c1f802a0bb7eb9ec4c185d07ff7d Mon Sep 17 00:00:00 2001 From: OliverB Date: Wed, 5 Aug 2020 10:50:12 +0200 Subject: [PATCH 55/63] Merged VecsConfigs --- fse/models/average_inner.c | 267 ++++++++++++++++------------------- fse/models/average_inner.pxd | 52 ++++--- fse/models/average_inner.pyx | 43 +++--- fse/models/pooling_inner.pyx | 38 +++-- 4 files changed, 191 insertions(+), 209 deletions(-) diff --git a/fse/models/average_inner.c b/fse/models/average_inner.c index 20349c1..8a1d75b 100644 --- a/fse/models/average_inner.c +++ b/fse/models/average_inner.c @@ -1132,8 +1132,7 @@ typedef npy_clongdouble __pyx_t_5numpy_clongdouble_t; * cdef inline object PyArray_MultiIterNew1(a): */ typedef npy_cdouble __pyx_t_5numpy_complex_t; -struct __pyx_t_13average_inner_BaseSentenceVecsConfig; -struct __pyx_t_13average_inner_FTSentenceVecsConfig; +struct __pyx_t_13average_inner_VecsConfig; /* "average_inner.pxd":20 * @@ -1156,11 +1155,11 @@ typedef void (*__pyx_t_13average_inner_sscal_ptr)(int const *, float const *, fl /* "average_inner.pxd":48 * DEF MAX_NGRAMS = 40 * - * cdef struct BaseSentenceVecsConfig: # <<<<<<<<<<<<<< + * cdef struct VecsConfig: # <<<<<<<<<<<<<< + * # Common * int size, workers - * */ -struct __pyx_t_13average_inner_BaseSentenceVecsConfig { +struct __pyx_t_13average_inner_VecsConfig { int size; int workers; __pyx_t_13average_inner_REAL_t *mem; @@ -1171,31 +1170,11 @@ struct __pyx_t_13average_inner_BaseSentenceVecsConfig { __pyx_t_13average_inner_uINT_t word_indices[0x2710]; __pyx_t_13average_inner_uINT_t sent_adresses[0x2710]; __pyx_t_13average_inner_uINT_t sentence_boundary[(0x2710 + 1)]; -}; - -/* "average_inner.pxd":62 - * uINT_t sentence_boundary[MAX_WORDS + 1] - * - * cdef struct FTSentenceVecsConfig: # <<<<<<<<<<<<<< - * int size, workers, min_n, max_n, bucket - * - */ -struct __pyx_t_13average_inner_FTSentenceVecsConfig { - int size; - int workers; int min_n; int max_n; int bucket; __pyx_t_13average_inner_REAL_t oov_weight; - __pyx_t_13average_inner_REAL_t *mem; - __pyx_t_13average_inner_REAL_t *mem2; - __pyx_t_13average_inner_REAL_t *word_vectors; __pyx_t_13average_inner_REAL_t *ngram_vectors; - __pyx_t_13average_inner_REAL_t *word_weights; - __pyx_t_13average_inner_REAL_t *sentence_vectors; - __pyx_t_13average_inner_uINT_t word_indices[0x2710]; - __pyx_t_13average_inner_uINT_t sent_adresses[0x2710]; - __pyx_t_13average_inner_uINT_t sentence_boundary[(0x2710 + 1)]; __pyx_t_13average_inner_uINT_t subwords_idx_len[0x2710]; __pyx_t_13average_inner_uINT_t *subwords_idx; }; @@ -1792,12 +1771,12 @@ static int __pyx_v_13average_inner_ONE; static int __pyx_v_13average_inner_ZERO; static __pyx_t_13average_inner_REAL_t __pyx_v_13average_inner_ONEF; static __pyx_t_13average_inner_REAL_t __pyx_v_13average_inner_ZEROF; -static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *, PyObject *); /*proto*/ -static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *, PyObject *); /*proto*/ -static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *); /*proto*/ -static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *); /*proto*/ -static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ -static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ +static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *, PyObject *); /*proto*/ +static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *, PyObject *); /*proto*/ +static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *); /*proto*/ +static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *); /*proto*/ +static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_t_13average_inner_VecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ +static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_13average_inner_VecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ #define __Pyx_MODULE_NAME "average_inner" extern int __pyx_module_is_main_average_inner; int __pyx_module_is_main_average_inner = 0; @@ -1809,13 +1788,11 @@ static PyObject *__pyx_builtin_ValueError; static PyObject *__pyx_builtin_RuntimeError; static PyObject *__pyx_builtin_ImportError; static const char __pyx_k__8[] = "*"; -static const char __pyx_k_ft[] = "ft"; static const char __pyx_k_np[] = "np"; static const char __pyx_k_sv[] = "sv"; static const char __pyx_k_wv[] = "wv"; static const char __pyx_k_ONE[] = "ONE"; static const char __pyx_k_max[] = "max"; -static const char __pyx_k_w2v[] = "w2v"; static const char __pyx_k_ONEF[] = "ONEF"; static const char __pyx_k_ZERO[] = "ZERO"; static const char __pyx_k_fill[] = "fill"; @@ -1836,6 +1813,7 @@ static const char __pyx_k_saxpy[] = "saxpy"; static const char __pyx_k_sscal[] = "sscal"; static const char __pyx_k_vocab[] = "vocab"; static const char __pyx_k_bucket[] = "bucket"; +static const char __pyx_k_config[] = "config"; static const char __pyx_k_import[] = "__import__"; static const char __pyx_k_memory[] = "memory"; static const char __pyx_k_target[] = "target"; @@ -1893,13 +1871,13 @@ static PyObject *__pyx_kp_s_average_inner_pyx; static PyObject *__pyx_n_s_bucket; static PyObject *__pyx_n_s_cline_in_traceback; static PyObject *__pyx_n_s_compute_ngrams_bytes; +static PyObject *__pyx_n_s_config; static PyObject *__pyx_n_s_cpointer; static PyObject *__pyx_n_s_eff_sentences; static PyObject *__pyx_n_s_eff_words; static PyObject *__pyx_n_s_enumerate; static PyObject *__pyx_n_s_fblas; static PyObject *__pyx_n_s_fill; -static PyObject *__pyx_n_s_ft; static PyObject *__pyx_n_s_ft_hash_bytes; static PyObject *__pyx_n_s_gensim_models__utils_any2vec; static PyObject *__pyx_n_s_import; @@ -1935,7 +1913,6 @@ static PyObject *__pyx_n_s_vectors; static PyObject *__pyx_n_s_vectors_ngrams; static PyObject *__pyx_n_s_vectors_vocab; static PyObject *__pyx_n_s_vocab; -static PyObject *__pyx_n_s_w2v; static PyObject *__pyx_n_s_word_weights; static PyObject *__pyx_n_s_workers; static PyObject *__pyx_n_s_wv; @@ -1963,11 +1940,11 @@ static PyObject *__pyx_codeobj__11; * DEF MAX_NGRAMS = 40 * * cdef init_base_s2v_config( # <<<<<<<<<<<<<< - * BaseSentenceVecsConfig *c, + * VecsConfig *c, * model, */ -static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, PyObject *__pyx_v_model, PyObject *__pyx_v_target, PyObject *__pyx_v_memory) { +static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13average_inner_VecsConfig *__pyx_v_c, PyObject *__pyx_v_model, PyObject *__pyx_v_target, PyObject *__pyx_v_memory) { PyObject *__pyx_r = NULL; __Pyx_RefNannyDeclarations PyObject *__pyx_t_1 = NULL; @@ -2073,7 +2050,7 @@ static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13a * DEF MAX_NGRAMS = 40 * * cdef init_base_s2v_config( # <<<<<<<<<<<<<< - * BaseSentenceVecsConfig *c, + * VecsConfig *c, * model, */ @@ -2095,11 +2072,11 @@ static PyObject *__pyx_f_13average_inner_init_base_s2v_config(struct __pyx_t_13a * c[0].sentence_vectors = (np.PyArray_DATA(target)) * * cdef init_ft_s2v_config( # <<<<<<<<<<<<<< - * FTSentenceVecsConfig *c, + * VecsConfig *c, * model, */ -static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, PyObject *__pyx_v_model, PyObject *__pyx_v_target, PyObject *__pyx_v_memory) { +static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13average_inner_VecsConfig *__pyx_v_c, PyObject *__pyx_v_model, PyObject *__pyx_v_target, PyObject *__pyx_v_memory) { PyObject *__pyx_r = NULL; __Pyx_RefNannyDeclarations PyObject *__pyx_t_1 = NULL; @@ -2351,7 +2328,7 @@ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13ave * c[0].sentence_vectors = (np.PyArray_DATA(target)) * * cdef init_ft_s2v_config( # <<<<<<<<<<<<<< - * FTSentenceVecsConfig *c, + * VecsConfig *c, * model, */ @@ -2375,11 +2352,11 @@ static PyObject *__pyx_f_13average_inner_init_ft_s2v_config(struct __pyx_t_13ave * c[0].sentence_vectors = (np.PyArray_DATA(target)) * * cdef object populate_base_s2v_config( # <<<<<<<<<<<<<< - * BaseSentenceVecsConfig *c, + * VecsConfig *c, * vocab, */ -static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, PyObject *__pyx_v_vocab, PyObject *__pyx_v_indexed_sentences) { +static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t_13average_inner_VecsConfig *__pyx_v_c, PyObject *__pyx_v_vocab, PyObject *__pyx_v_indexed_sentences) { __pyx_t_13average_inner_uINT_t __pyx_v_eff_words; __pyx_t_13average_inner_uINT_t __pyx_v_eff_sents; PyObject *__pyx_v_obj = NULL; @@ -2770,7 +2747,7 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t * c[0].sentence_vectors = (np.PyArray_DATA(target)) * * cdef object populate_base_s2v_config( # <<<<<<<<<<<<<< - * BaseSentenceVecsConfig *c, + * VecsConfig *c, * vocab, */ @@ -2795,11 +2772,11 @@ static PyObject *__pyx_f_13average_inner_populate_base_s2v_config(struct __pyx_t * return eff_sents, eff_words * * cdef object populate_ft_s2v_config( # <<<<<<<<<<<<<< - * FTSentenceVecsConfig *c, + * VecsConfig *c, * vocab, */ -static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, PyObject *__pyx_v_vocab, PyObject *__pyx_v_indexed_sentences) { +static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_13average_inner_VecsConfig *__pyx_v_c, PyObject *__pyx_v_vocab, PyObject *__pyx_v_indexed_sentences) { __pyx_t_13average_inner_uINT_t __pyx_v_eff_words; __pyx_t_13average_inner_uINT_t __pyx_v_eff_sents; PyObject *__pyx_v_obj = NULL; @@ -3431,7 +3408,7 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 * return eff_sents, eff_words * * cdef object populate_ft_s2v_config( # <<<<<<<<<<<<<< - * FTSentenceVecsConfig *c, + * VecsConfig *c, * vocab, */ @@ -3465,11 +3442,11 @@ static PyObject *__pyx_f_13average_inner_populate_ft_s2v_config(struct __pyx_t_1 * * * cdef void compute_base_sentence_averages( # <<<<<<<<<<<<<< - * BaseSentenceVecsConfig *c, + * VecsConfig *c, * uINT_t num_sentences, */ -static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { +static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_t_13average_inner_VecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { int __pyx_v_size; __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; @@ -3647,7 +3624,7 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ * * * cdef void compute_base_sentence_averages( # <<<<<<<<<<<<<< - * BaseSentenceVecsConfig *c, + * VecsConfig *c, * uINT_t num_sentences, */ @@ -3658,11 +3635,11 @@ static void __pyx_f_13average_inner_compute_base_sentence_averages(struct __pyx_ * ) * * cdef void compute_ft_sentence_averages( # <<<<<<<<<<<<<< - * FTSentenceVecsConfig *c, + * VecsConfig *c, * uINT_t num_sentences, */ -static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { +static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_13average_inner_VecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { int __pyx_v_size; __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; @@ -3931,7 +3908,7 @@ static void __pyx_f_13average_inner_compute_ft_sentence_averages(struct __pyx_t_ * ) * * cdef void compute_ft_sentence_averages( # <<<<<<<<<<<<<< - * FTSentenceVecsConfig *c, + * VecsConfig *c, * uINT_t num_sentences, */ @@ -4034,8 +4011,7 @@ static PyObject *__pyx_pw_13average_inner_1train_average_cy(PyObject *__pyx_self static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObject *__pyx_self, PyObject *__pyx_v_model, PyObject *__pyx_v_indexed_sentences, PyObject *__pyx_v_target, PyObject *__pyx_v_memory) { __pyx_t_13average_inner_uINT_t __pyx_v_eff_sentences; __pyx_t_13average_inner_uINT_t __pyx_v_eff_words; - struct __pyx_t_13average_inner_BaseSentenceVecsConfig __pyx_v_w2v; - struct __pyx_t_13average_inner_FTSentenceVecsConfig __pyx_v_ft; + struct __pyx_t_13average_inner_VecsConfig __pyx_v_config; PyObject *__pyx_r = NULL; __Pyx_RefNannyDeclarations PyObject *__pyx_t_1 = NULL; @@ -4054,7 +4030,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec * * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< * cdef uINT_t eff_words = 0 - * cdef BaseSentenceVecsConfig w2v + * cdef VecsConfig config */ __pyx_v_eff_sentences = 0; @@ -4062,57 +4038,57 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec * * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< - * cdef BaseSentenceVecsConfig w2v - * cdef FTSentenceVecsConfig ft + * cdef VecsConfig config + * */ __pyx_v_eff_words = 0; - /* "average_inner.pyx":414 - * cdef FTSentenceVecsConfig ft + /* "average_inner.pyx":413 + * cdef VecsConfig config * * if not model.is_ft: # <<<<<<<<<<<<<< - * init_base_s2v_config(&w2v, model, target, memory) + * init_base_s2v_config(&config, model, target, memory) * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 414, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 413, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 414, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) __PYX_ERR(0, 413, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_3 = ((!__pyx_t_2) != 0); if (__pyx_t_3) { - /* "average_inner.pyx":415 + /* "average_inner.pyx":414 * * if not model.is_ft: - * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< + * init_base_s2v_config(&config, model, target, memory) # <<<<<<<<<<<<<< * * eff_sentences, eff_words = populate_base_s2v_config( */ - __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 415, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_config), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 414, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":419 + /* "average_inner.pyx":418 * eff_sentences, eff_words = populate_base_s2v_config( - * &w2v, + * &config, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences, * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 419, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 418, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 419, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 418, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":417 - * init_base_s2v_config(&w2v, model, target, memory) + /* "average_inner.pyx":416 + * init_base_s2v_config(&config, model, target, memory) * * eff_sentences, eff_words = populate_base_s2v_config( # <<<<<<<<<<<<<< - * &w2v, + * &config, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 417, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_config), __pyx_t_4, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 416, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -4121,7 +4097,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 417, __pyx_L1_error) + __PYX_ERR(0, 416, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -4134,15 +4110,15 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_4); __Pyx_INCREF(__pyx_t_5); #else - __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 417, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 416, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 417, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 416, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 417, __pyx_L1_error) + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 416, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; @@ -4150,7 +4126,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_4); index = 1; __pyx_t_5 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_5)) goto __pyx_L4_unpacking_failed; __Pyx_GOTREF(__pyx_t_5); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 417, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 416, __pyx_L1_error) __pyx_t_7 = NULL; __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; goto __pyx_L5_unpacking_done; @@ -4158,22 +4134,22 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_t_7 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 417, __pyx_L1_error) + __PYX_ERR(0, 416, __pyx_L1_error) __pyx_L5_unpacking_done:; } - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 417, __pyx_L1_error) + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 416, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 417, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 416, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; __pyx_v_eff_sentences = __pyx_t_8; __pyx_v_eff_words = __pyx_t_9; - /* "average_inner.pyx":423 + /* "average_inner.pyx":422 * ) * * with nogil: # <<<<<<<<<<<<<< * compute_base_sentence_averages( - * &w2v, + * &config, */ { #ifdef WITH_THREAD @@ -4183,22 +4159,22 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "average_inner.pyx":424 + /* "average_inner.pyx":423 * * with nogil: * compute_base_sentence_averages( # <<<<<<<<<<<<<< - * &w2v, + * &config, * eff_sentences, */ - __pyx_f_13average_inner_compute_base_sentence_averages((&__pyx_v_w2v), __pyx_v_eff_sentences); + __pyx_f_13average_inner_compute_base_sentence_averages((&__pyx_v_config), __pyx_v_eff_sentences); } - /* "average_inner.pyx":423 + /* "average_inner.pyx":422 * ) * * with nogil: # <<<<<<<<<<<<<< * compute_base_sentence_averages( - * &w2v, + * &config, */ /*finally:*/ { /*normal exit:*/{ @@ -4212,57 +4188,57 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec } } - /* "average_inner.pyx":414 - * cdef FTSentenceVecsConfig ft + /* "average_inner.pyx":413 + * cdef VecsConfig config * * if not model.is_ft: # <<<<<<<<<<<<<< - * init_base_s2v_config(&w2v, model, target, memory) + * init_base_s2v_config(&config, model, target, memory) * */ goto __pyx_L3; } - /* "average_inner.pyx":429 + /* "average_inner.pyx":428 * ) * else: * init_ft_s2v_config( # <<<<<<<<<<<<<< - * &ft, + * &config, * model, */ /*else*/ { - /* "average_inner.pyx":433 + /* "average_inner.pyx":432 * model, * target, * memory, # <<<<<<<<<<<<<< * ) * */ - __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 429, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_config), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 428, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":438 + /* "average_inner.pyx":437 * eff_sentences, eff_words = populate_ft_s2v_config( - * &ft, + * &config, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences, * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 438, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 437, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 438, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 437, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":436 + /* "average_inner.pyx":435 * ) * * eff_sentences, eff_words = populate_ft_s2v_config( # <<<<<<<<<<<<<< - * &ft, + * &config, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 436, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_config), __pyx_t_5, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 435, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -4271,7 +4247,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 436, __pyx_L1_error) + __PYX_ERR(0, 435, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -4284,15 +4260,15 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_5); __Pyx_INCREF(__pyx_t_4); #else - __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 436, __pyx_L1_error) + __pyx_t_5 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 435, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); - __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 436, __pyx_L1_error) + __pyx_t_4 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 435, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 436, __pyx_L1_error) + __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 435, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_7 = Py_TYPE(__pyx_t_6)->tp_iternext; @@ -4300,7 +4276,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_5); index = 1; __pyx_t_4 = __pyx_t_7(__pyx_t_6); if (unlikely(!__pyx_t_4)) goto __pyx_L9_unpacking_failed; __Pyx_GOTREF(__pyx_t_4); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 436, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_7(__pyx_t_6), 2) < 0) __PYX_ERR(0, 435, __pyx_L1_error) __pyx_t_7 = NULL; __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; goto __pyx_L10_unpacking_done; @@ -4308,22 +4284,22 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_t_7 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 436, __pyx_L1_error) + __PYX_ERR(0, 435, __pyx_L1_error) __pyx_L10_unpacking_done:; } - __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 436, __pyx_L1_error) + __pyx_t_9 = __Pyx_PyInt_As_npy_uint32(__pyx_t_5); if (unlikely((__pyx_t_9 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 435, __pyx_L1_error) __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0; - __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 436, __pyx_L1_error) + __pyx_t_8 = __Pyx_PyInt_As_npy_uint32(__pyx_t_4); if (unlikely((__pyx_t_8 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 435, __pyx_L1_error) __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0; __pyx_v_eff_sentences = __pyx_t_9; __pyx_v_eff_words = __pyx_t_8; - /* "average_inner.pyx":442 + /* "average_inner.pyx":441 * ) * * with nogil: # <<<<<<<<<<<<<< * compute_ft_sentence_averages( - * &ft, + * &config, */ { #ifdef WITH_THREAD @@ -4333,22 +4309,22 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "average_inner.pyx":443 + /* "average_inner.pyx":442 * * with nogil: * compute_ft_sentence_averages( # <<<<<<<<<<<<<< - * &ft, + * &config, * eff_sentences */ - __pyx_f_13average_inner_compute_ft_sentence_averages((&__pyx_v_ft), __pyx_v_eff_sentences); + __pyx_f_13average_inner_compute_ft_sentence_averages((&__pyx_v_config), __pyx_v_eff_sentences); } - /* "average_inner.pyx":442 + /* "average_inner.pyx":441 * ) * * with nogil: # <<<<<<<<<<<<<< * compute_ft_sentence_averages( - * &ft, + * &config, */ /*finally:*/ { /*normal exit:*/{ @@ -4364,7 +4340,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec } __pyx_L3:; - /* "average_inner.pyx":448 + /* "average_inner.pyx":447 * ) * * return eff_sentences, eff_words # <<<<<<<<<<<<<< @@ -4372,11 +4348,11 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec * def init(): */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 448, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 447, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 448, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_4)) __PYX_ERR(0, 447, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_4); - __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 448, __pyx_L1_error) + __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) __PYX_ERR(0, 447, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_5); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_1); @@ -4410,7 +4386,7 @@ static PyObject *__pyx_pf_13average_inner_train_average_cy(CYTHON_UNUSED PyObjec return __pyx_r; } -/* "average_inner.pyx":450 +/* "average_inner.pyx":449 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -4438,7 +4414,7 @@ static PyObject *__pyx_pf_13average_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("init", 0); - /* "average_inner.pyx":451 + /* "average_inner.pyx":450 * * def init(): * return 1 # <<<<<<<<<<<<<< @@ -4450,7 +4426,7 @@ static PyObject *__pyx_pf_13average_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __pyx_r = __pyx_int_1; goto __pyx_L0; - /* "average_inner.pyx":450 + /* "average_inner.pyx":449 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -6948,13 +6924,13 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = { {&__pyx_n_s_bucket, __pyx_k_bucket, sizeof(__pyx_k_bucket), 0, 0, 1, 1}, {&__pyx_n_s_cline_in_traceback, __pyx_k_cline_in_traceback, sizeof(__pyx_k_cline_in_traceback), 0, 0, 1, 1}, {&__pyx_n_s_compute_ngrams_bytes, __pyx_k_compute_ngrams_bytes, sizeof(__pyx_k_compute_ngrams_bytes), 0, 0, 1, 1}, + {&__pyx_n_s_config, __pyx_k_config, sizeof(__pyx_k_config), 0, 0, 1, 1}, {&__pyx_n_s_cpointer, __pyx_k_cpointer, sizeof(__pyx_k_cpointer), 0, 0, 1, 1}, {&__pyx_n_s_eff_sentences, __pyx_k_eff_sentences, sizeof(__pyx_k_eff_sentences), 0, 0, 1, 1}, {&__pyx_n_s_eff_words, __pyx_k_eff_words, sizeof(__pyx_k_eff_words), 0, 0, 1, 1}, {&__pyx_n_s_enumerate, __pyx_k_enumerate, sizeof(__pyx_k_enumerate), 0, 0, 1, 1}, {&__pyx_n_s_fblas, __pyx_k_fblas, sizeof(__pyx_k_fblas), 0, 0, 1, 1}, {&__pyx_n_s_fill, __pyx_k_fill, sizeof(__pyx_k_fill), 0, 0, 1, 1}, - {&__pyx_n_s_ft, __pyx_k_ft, sizeof(__pyx_k_ft), 0, 0, 1, 1}, {&__pyx_n_s_ft_hash_bytes, __pyx_k_ft_hash_bytes, sizeof(__pyx_k_ft_hash_bytes), 0, 0, 1, 1}, {&__pyx_n_s_gensim_models__utils_any2vec, __pyx_k_gensim_models__utils_any2vec, sizeof(__pyx_k_gensim_models__utils_any2vec), 0, 0, 1, 1}, {&__pyx_n_s_import, __pyx_k_import, sizeof(__pyx_k_import), 0, 0, 1, 1}, @@ -6990,7 +6966,6 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = { {&__pyx_n_s_vectors_ngrams, __pyx_k_vectors_ngrams, sizeof(__pyx_k_vectors_ngrams), 0, 0, 1, 1}, {&__pyx_n_s_vectors_vocab, __pyx_k_vectors_vocab, sizeof(__pyx_k_vectors_vocab), 0, 0, 1, 1}, {&__pyx_n_s_vocab, __pyx_k_vocab, sizeof(__pyx_k_vocab), 0, 0, 1, 1}, - {&__pyx_n_s_w2v, __pyx_k_w2v, sizeof(__pyx_k_w2v), 0, 0, 1, 1}, {&__pyx_n_s_word_weights, __pyx_k_word_weights, sizeof(__pyx_k_word_weights), 0, 0, 1, 1}, {&__pyx_n_s_workers, __pyx_k_workers, sizeof(__pyx_k_workers), 0, 0, 1, 1}, {&__pyx_n_s_wv, __pyx_k_wv, sizeof(__pyx_k_wv), 0, 0, 1, 1}, @@ -7095,19 +7070,19 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { * model, * indexed_sentences, */ - __pyx_tuple__9 = PyTuple_Pack(8, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 380, __pyx_L1_error) + __pyx_tuple__9 = PyTuple_Pack(7, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_config); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 380, __pyx_L1_error) __Pyx_GOTREF(__pyx_tuple__9); __Pyx_GIVEREF(__pyx_tuple__9); - __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 8, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_train_average_cy, 380, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 380, __pyx_L1_error) + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 7, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_train_average_cy, 380, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 380, __pyx_L1_error) - /* "average_inner.pyx":450 + /* "average_inner.pyx":449 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_init, 450, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 450, __pyx_L1_error) + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_average_inner_pyx, __pyx_n_s_init, 449, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 449, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -7163,10 +7138,10 @@ static int __Pyx_modinit_function_export_code(void) { __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("__Pyx_modinit_function_export_code", 0); /*--- Function export code ---*/ - if (__Pyx_ExportFunction("init_base_s2v_config", (void (*)(void))__pyx_f_13average_inner_init_base_s2v_config, "PyObject *(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) - if (__Pyx_ExportFunction("init_ft_s2v_config", (void (*)(void))__pyx_f_13average_inner_init_ft_s2v_config, "PyObject *(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) - if (__Pyx_ExportFunction("populate_base_s2v_config", (void (*)(void))__pyx_f_13average_inner_populate_base_s2v_config, "PyObject *(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) - if (__Pyx_ExportFunction("populate_ft_s2v_config", (void (*)(void))__pyx_f_13average_inner_populate_ft_s2v_config, "PyObject *(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ExportFunction("init_base_s2v_config", (void (*)(void))__pyx_f_13average_inner_init_base_s2v_config, "PyObject *(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ExportFunction("init_ft_s2v_config", (void (*)(void))__pyx_f_13average_inner_init_ft_s2v_config, "PyObject *(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ExportFunction("populate_base_s2v_config", (void (*)(void))__pyx_f_13average_inner_populate_base_s2v_config, "PyObject *(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ExportFunction("populate_ft_s2v_config", (void (*)(void))__pyx_f_13average_inner_populate_ft_s2v_config, "PyObject *(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -7575,46 +7550,46 @@ if (!__Pyx_RefNanny) { if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_average_cy, __pyx_t_1) < 0) __PYX_ERR(0, 380, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":450 + /* "average_inner.pyx":449 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_3init, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 450, __pyx_L1_error) + __pyx_t_1 = PyCFunction_NewEx(&__pyx_mdef_13average_inner_3init, NULL, __pyx_n_s_average_inner); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 449, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_1) < 0) __PYX_ERR(0, 450, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_1) < 0) __PYX_ERR(0, 449, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "average_inner.pyx":453 + /* "average_inner.pyx":452 * return 1 * * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 453, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 452, __pyx_L1_error) - /* "average_inner.pyx":454 + /* "average_inner.pyx":453 * * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 454, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 453, __pyx_L1_error) - /* "average_inner.pyx":455 + /* "average_inner.pyx":454 * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() # <<<<<<<<<<<<<< */ - __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_init); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 455, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_1, __pyx_n_s_init); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 454, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyObject_CallNoArg(__pyx_t_1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 455, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyObject_CallNoArg(__pyx_t_1); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 454, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_2) < 0) __PYX_ERR(0, 455, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_2) < 0) __PYX_ERR(0, 454, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; /* "average_inner.pyx":1 diff --git a/fse/models/average_inner.pxd b/fse/models/average_inner.pxd index f5e984a..326cf9e 100644 --- a/fse/models/average_inner.pxd +++ b/fse/models/average_inner.pxd @@ -45,10 +45,10 @@ cdef REAL_t ZEROF = 0.0 DEF MAX_WORDS = 10000 DEF MAX_NGRAMS = 40 -cdef struct BaseSentenceVecsConfig: +cdef struct VecsConfig: + # Common int size, workers - # Vectors REAL_t *mem REAL_t *mem2 REAL_t *word_vectors @@ -59,49 +59,59 @@ cdef struct BaseSentenceVecsConfig: uINT_t sent_adresses[MAX_WORDS] uINT_t sentence_boundary[MAX_WORDS + 1] -cdef struct FTSentenceVecsConfig: - int size, workers, min_n, max_n, bucket + # Fasttext specific + int min_n, max_n, bucket REAL_t oov_weight - - # Vectors - REAL_t *mem - REAL_t *mem2 - REAL_t *word_vectors # Note: these will be the vocab vectors, not wv.vectors REAL_t *ngram_vectors - REAL_t *word_weights - - REAL_t *sentence_vectors - - # REAL_t *work memory for summation? - uINT_t word_indices[MAX_WORDS] - uINT_t sent_adresses[MAX_WORDS] - uINT_t sentence_boundary[MAX_WORDS + 1] # For storing the oov items uINT_t subwords_idx_len[MAX_WORDS] uINT_t *subwords_idx + +# cdef struct FTSentenceVecsConfig: +# int size, workers, min_n, max_n, bucket + +# REAL_t oov_weight + +# # Vectors +# REAL_t *mem +# REAL_t *mem2 +# REAL_t *word_vectors # Note: these will be the vocab vectors, not wv.vectors +# REAL_t *ngram_vectors +# REAL_t *word_weights + +# REAL_t *sentence_vectors + +# # REAL_t *work memory for summation? +# uINT_t word_indices[MAX_WORDS] +# uINT_t sent_adresses[MAX_WORDS] +# uINT_t sentence_boundary[MAX_WORDS + 1] + +# # For storing the oov items +# uINT_t subwords_idx_len[MAX_WORDS] +# uINT_t *subwords_idx cdef init_base_s2v_config( - BaseSentenceVecsConfig *c, + VecsConfig *c, model, target, memory ) cdef init_ft_s2v_config( - FTSentenceVecsConfig *c, + VecsConfig *c, model, target, memory ) cdef object populate_base_s2v_config( - BaseSentenceVecsConfig *c, + VecsConfig *c, vocab, indexed_sentences, ) cdef object populate_ft_s2v_config( - FTSentenceVecsConfig *c, + VecsConfig *c, vocab, indexed_sentences, ) \ No newline at end of file diff --git a/fse/models/average_inner.pyx b/fse/models/average_inner.pyx index 09820e5..cbaec06 100644 --- a/fse/models/average_inner.pyx +++ b/fse/models/average_inner.pyx @@ -36,16 +36,16 @@ DEF MAX_WORDS = 10000 DEF MAX_NGRAMS = 40 cdef init_base_s2v_config( - BaseSentenceVecsConfig *c, + VecsConfig *c, model, target, memory, ): - """Load BaseAny2Vec parameters into a BaseSentenceVecsConfig struct. + """Load BaseAny2Vec parameters into a VecsConfig struct. Parameters ---------- - c : FTSentenceVecsConfig * + c : VecsConfig * A pointer to the struct to initialize. model : fse.models.base_s2v.BaseSentence2VecModel The model to load. @@ -68,16 +68,16 @@ cdef init_base_s2v_config( c[0].sentence_vectors = (np.PyArray_DATA(target)) cdef init_ft_s2v_config( - FTSentenceVecsConfig *c, + VecsConfig *c, model, target, memory, ): - """Load Fasttext parameters into a FTSentenceVecsConfig struct. + """Load Fasttext parameters into a VecsConfig struct. Parameters ---------- - c : FTSentenceVecsConfig * + c : VecsConfig * A pointer to the struct to initialize. model : fse.models.base_s2v.BaseSentence2VecModel The model to load. @@ -110,7 +110,7 @@ cdef init_ft_s2v_config( c[0].sentence_vectors = (np.PyArray_DATA(target)) cdef object populate_base_s2v_config( - BaseSentenceVecsConfig *c, + VecsConfig *c, vocab, indexed_sentences, ): @@ -121,7 +121,7 @@ cdef object populate_base_s2v_config( Parameters ---------- - c : BaseSentenceVecsConfig* + c : VecsConfig* A pointer to the struct that will contain the populated indices. vocab : dict The vocabulary @@ -164,7 +164,7 @@ cdef object populate_base_s2v_config( return eff_sents, eff_words cdef object populate_ft_s2v_config( - FTSentenceVecsConfig *c, + VecsConfig *c, vocab, indexed_sentences, ): @@ -175,7 +175,7 @@ cdef object populate_ft_s2v_config( Parameters ---------- - c : FTSentenceVecsConfig* + c : VecsConfig* A pointer to the struct that will contain the populated indices. vocab : dict The vocabulary @@ -235,14 +235,14 @@ cdef object populate_ft_s2v_config( cdef void compute_base_sentence_averages( - BaseSentenceVecsConfig *c, + VecsConfig *c, uINT_t num_sentences, ) nogil: """Perform optimized sentence-level averaging for BaseAny2Vec model. Parameters ---------- - c : BaseSentenceVecsConfig * + c : VecsConfig * A pointer to a fully initialized and populated struct. num_sentences : uINT_t The number of sentences used to train the model. @@ -299,14 +299,14 @@ cdef void compute_base_sentence_averages( ) cdef void compute_ft_sentence_averages( - FTSentenceVecsConfig *c, + VecsConfig *c, uINT_t num_sentences, ) nogil: """Perform optimized sentence-level averaging for FastText model. Parameters ---------- - c : FTSentenceVecsConfig * + c : VecsConfig * A pointer to a fully initialized and populated struct. num_sentences : uINT_t The number of sentences used to train the model. @@ -408,40 +408,39 @@ def train_average_cy( cdef uINT_t eff_sentences = 0 cdef uINT_t eff_words = 0 - cdef BaseSentenceVecsConfig w2v - cdef FTSentenceVecsConfig ft + cdef VecsConfig config if not model.is_ft: - init_base_s2v_config(&w2v, model, target, memory) + init_base_s2v_config(&config, model, target, memory) eff_sentences, eff_words = populate_base_s2v_config( - &w2v, + &config, model.wv.vocab, indexed_sentences, ) with nogil: compute_base_sentence_averages( - &w2v, + &config, eff_sentences, ) else: init_ft_s2v_config( - &ft, + &config, model, target, memory, ) eff_sentences, eff_words = populate_ft_s2v_config( - &ft, + &config, model.wv.vocab, indexed_sentences, ) with nogil: compute_ft_sentence_averages( - &ft, + &config, eff_sentences ) diff --git a/fse/models/pooling_inner.pyx b/fse/models/pooling_inner.pyx index 54e9c0a..3878735 100644 --- a/fse/models/pooling_inner.pyx +++ b/fse/models/pooling_inner.pyx @@ -30,8 +30,7 @@ from average_inner cimport ( ZEROF, saxpy, sscal, - BaseSentenceVecsConfig, - FTSentenceVecsConfig, + VecsConfig, init_base_s2v_config, init_ft_s2v_config, populate_base_s2v_config, @@ -67,14 +66,14 @@ cdef void swrmax_pool( Y[i] = alpha[0] * X[i] cdef void compute_base_sentence_pooling( - BaseSentenceVecsConfig *c, + VecsConfig *c, uINT_t num_sentences, ) nogil: """Perform optimized sentence-level max pooling for BaseAny2Vec model. Parameters ---------- - c : BaseSentenceVecsConfig * + c : VecsConfig * A pointer to a fully initialized and populated struct. num_sentences : uINT_t The number of sentences used to train the model. @@ -114,7 +113,7 @@ cdef void compute_base_sentence_pooling( cdef void compute_base_sentence_hier_pooling( - BaseSentenceVecsConfig *c, + VecsConfig *c, uINT_t num_sentences, uINT_t window_size, REAL_t window_stride, @@ -123,7 +122,7 @@ cdef void compute_base_sentence_hier_pooling( Parameters ---------- - c : BaseSentenceVecsConfig * + c : VecsConfig * A pointer to a fully initialized and populated struct. num_sentences : uINT_t The number of sentences used to train the model. @@ -203,14 +202,14 @@ cdef void compute_base_sentence_hier_pooling( ) cdef void compute_ft_sentence_pooling( - FTSentenceVecsConfig *c, + VecsConfig *c, uINT_t num_sentences, ) nogil: """Perform optimized sentence-level max pooling for FastText model. Parameters ---------- - c : FTSentenceVecsConfig * + c : VecsConfig * A pointer to a fully initialized and populated struct. num_sentences : uINT_t The number of sentences used to train the model. @@ -278,7 +277,7 @@ cdef void compute_ft_sentence_pooling( # There's nothing to do here for many-to-one mappings cdef void compute_ft_sentence_hier_pooling( - FTSentenceVecsConfig *c, + VecsConfig *c, uINT_t num_sentences, uINT_t window_size, REAL_t window_stride, @@ -287,7 +286,7 @@ cdef void compute_ft_sentence_hier_pooling( Parameters ---------- - c : FTSentenceVecsConfig * + c : VecsConfig * A pointer to a fully initialized and populated struct. num_sentences : uINT_t The number of sentences used to train the model. @@ -431,14 +430,13 @@ def train_pooling_cy( cdef uINT_t eff_words = 0 cdef uINT_t window_size = model.window_size cdef REAL_t window_stride = model.window_stride - cdef BaseSentenceVecsConfig w2v - cdef FTSentenceVecsConfig ft + cdef VecsConfig config if not model.is_ft: - init_base_s2v_config(&w2v, model, target, memory) + init_base_s2v_config(&config, model, target, memory) eff_sentences, eff_words = populate_base_s2v_config( - &w2v, + &config, model.wv.vocab, indexed_sentences ) @@ -446,22 +444,22 @@ def train_pooling_cy( if not model.hierarchical: with nogil: compute_base_sentence_pooling( - &w2v, + &config, eff_sentences ) else: with nogil: compute_base_sentence_hier_pooling( - &w2v, + &config, eff_sentences, window_size, window_stride, ) else: - init_ft_s2v_config(&ft, model, target, memory) + init_ft_s2v_config(&config, model, target, memory) eff_sentences, eff_words = populate_ft_s2v_config( - &ft, + &config, model.wv.vocab, indexed_sentences ) @@ -469,13 +467,13 @@ def train_pooling_cy( if not model.hierarchical: with nogil: compute_ft_sentence_pooling( - &ft, + &config, eff_sentences ) else: with nogil: compute_ft_sentence_hier_pooling( - &ft, + &config, eff_sentences, window_size, window_stride, From 67192c11161fcf7ea9a9ed73b26f45967be596b2 Mon Sep 17 00:00:00 2001 From: OliverB Date: Wed, 5 Aug 2020 10:50:21 +0200 Subject: [PATCH 56/63] Bugfix --- fse/models/base_iterator.py | 195 ++++++++++++ fse/models/pooling_inner.c | 569 +++++++++++++++++------------------- fse/test/test_iterator.py | 43 +++ 3 files changed, 510 insertions(+), 297 deletions(-) create mode 100644 fse/models/base_iterator.py create mode 100644 fse/test/test_iterator.py diff --git a/fse/models/base_iterator.py b/fse/models/base_iterator.py new file mode 100644 index 0000000..8a18e9e --- /dev/null +++ b/fse/models/base_iterator.py @@ -0,0 +1,195 @@ +#!/usr/bin/env python +# -*- coding: utf-8 -*- + +# Author: Oliver Borchers +# Copyright (C) 2020 Oliver Borchers +# For License information, see corresponding LICENSE file. + +from fse.models.utils import get_ft_word_vector +from typing import List + +import logging + +logger = logging.getLogger(__name__) + +from numpy import( + ndarray, + zeros, + ceil, + float32 as REAL, + amax as np_amax, + max as np_max, +) + +def base_iterator( + model, + indexed_sentences: List[tuple], + target: ndarray, + memory: tuple, + window_merger : callable, + sentence_merger : callable, +) -> [int, int]: + # """Training on a sequence of sentences and update the target ndarray. + + # Called internally from :meth:`~fse.models.pooling.MaxPooling._do_train_job`. + + # Warnings + # -------- + # This is the non-optimized, pure Python version. If you have a C compiler, + # fse will use an optimized code path from :mod:`fse.models.pooling_inner` instead. + + # Parameters + # ---------- + # model : :class:`~fse.models.base_s2v.BaseSentence2VecModel` + # The BaseSentence2VecModel model instance. + # indexed_sentences : iterable of tuple + # The sentences used to train the model. + # target : ndarray + # The target ndarray. We use the index from indexed_sentences + # to write into the corresponding row of target. + # memory : tuple + # Private memory array(s) for each working thread + + # Returns + # ------- + # int, int + # Number of effective sentences (non-zero) and effective words in the vocabulary used + # during training the sentence embedding. + + # """ + size = model.wv.vector_size + vocab = model.wv.vocab + + w_vectors = model.wv.vectors + w_weights = model.word_weights + + s_vectors = target + + is_ft = model.is_ft + + mem = memory[0] + mem2 = memory[1] + + window_size = model.window_size + window_stride = model.window_stride + + if is_ft: + # NOTE: For Fasttext: Use wv.vectors_vocab + # Using the wv.vectors from fasttext had horrible effects on the sts results + # I suspect this is because the wv.vectors are based on the averages of + # wv.vectors_vocab + wv.vectors_ngrams, which will all point into very + # similar directions. + max_ngrams = model.batch_ngrams + w_vectors = model.wv.vectors_vocab + ngram_vectors = model.wv.vectors_ngrams + min_n = model.wv.min_n + max_n = model.wv.max_n + bucket = model.wv.bucket + oov_weight = np_amax(w_weights) + + eff_sentences, eff_words = 0, 0 + + for obj in indexed_sentences: + sent = obj[0] + sent_adr = obj[1] + + if not len(sent): + continue + + eff_sentences += 1 + + # In cython we know the length (-OOV) beforehand + sent_len = sentence_length( + model, + sent, + ) + if not sent_len: + continue + + # Number of windows to be encountered + win_count = int(ceil(sent_len / window_size)) + + for word_index, _ in enumerate(sent): + if word_index % window_stride != 0: + continue + + win_len = 0 + mem.fill(0.) # mem for summation + mem2.fill(0.) + for word in sent[word_index : word_index + window_size]: + eff_words += window_func( + model, + word, + mem, + ) # -> mem + # W2V will return 0&1, FT returns 1 + + win_len += 1 + + # Herein the window will be merged (i.e., rescaled) + window_merger( + model, + win_len, + mem, + mem2, + ) # mem -> mem2 + + # Partially constructs the sentence onto sv.vectors + sentence_merger( + sent_len, + mem2, + sent_adr, + target, + ) + + return eff_sentences, eff_words + +def sentence_length( + model, + sent : List, + ) -> int: + """ We know the length of the sentence a-priori + """ + if model.is_ft: + return len(sent) + else: + # Inefficient, but hey, its just the python version anyways + return sum([1 if token in model.wv.vocab else 0 for token in sent]) + +def window_func( + model, + word : str, + mem : ndarray, + ) -> int: + """ Computes the word vectors for a word + """ + if word in model.wv.vocab: + word_index = model.wv.vocab[word].index + mem += model.wv.vectors[word_index] * model.word_weights[word_index] + return 1 + else: + if model.is_ft: + mem += get_ft_word_vector(word, model) * np_max(model.word_weights) + return 1 + else: + return 0 # Word not taken into account + +def window_merger( + model, + win_len : int, + mem : ndarray, + mem2 : ndarray, + ): + """ Average window merger. + Should implement functionality to merge temporary results from + mem to mem2 inplace. Depends on model architecture + """ + pass + +def sentence_merger( + window_length : int, + mem : ndarray, + sent_adr : int, + target : ndarray, + ): + pass \ No newline at end of file diff --git a/fse/models/pooling_inner.c b/fse/models/pooling_inner.c index e67daa0..00341f8 100644 --- a/fse/models/pooling_inner.c +++ b/fse/models/pooling_inner.c @@ -1135,8 +1135,7 @@ typedef npy_clongdouble __pyx_t_5numpy_clongdouble_t; * cdef inline object PyArray_MultiIterNew1(a): */ typedef npy_cdouble __pyx_t_5numpy_complex_t; -struct __pyx_t_13average_inner_BaseSentenceVecsConfig; -struct __pyx_t_13average_inner_FTSentenceVecsConfig; +struct __pyx_t_13average_inner_VecsConfig; /* "average_inner.pxd":20 * @@ -1159,11 +1158,11 @@ typedef void (*__pyx_t_13average_inner_sscal_ptr)(int const *, float const *, fl /* "average_inner.pxd":48 * DEF MAX_NGRAMS = 40 * - * cdef struct BaseSentenceVecsConfig: # <<<<<<<<<<<<<< + * cdef struct VecsConfig: # <<<<<<<<<<<<<< + * # Common * int size, workers - * */ -struct __pyx_t_13average_inner_BaseSentenceVecsConfig { +struct __pyx_t_13average_inner_VecsConfig { int size; int workers; __pyx_t_13average_inner_REAL_t *mem; @@ -1174,31 +1173,11 @@ struct __pyx_t_13average_inner_BaseSentenceVecsConfig { __pyx_t_13average_inner_uINT_t word_indices[0x2710]; __pyx_t_13average_inner_uINT_t sent_adresses[0x2710]; __pyx_t_13average_inner_uINT_t sentence_boundary[(0x2710 + 1)]; -}; - -/* "average_inner.pxd":62 - * uINT_t sentence_boundary[MAX_WORDS + 1] - * - * cdef struct FTSentenceVecsConfig: # <<<<<<<<<<<<<< - * int size, workers, min_n, max_n, bucket - * - */ -struct __pyx_t_13average_inner_FTSentenceVecsConfig { - int size; - int workers; int min_n; int max_n; int bucket; __pyx_t_13average_inner_REAL_t oov_weight; - __pyx_t_13average_inner_REAL_t *mem; - __pyx_t_13average_inner_REAL_t *mem2; - __pyx_t_13average_inner_REAL_t *word_vectors; __pyx_t_13average_inner_REAL_t *ngram_vectors; - __pyx_t_13average_inner_REAL_t *word_weights; - __pyx_t_13average_inner_REAL_t *sentence_vectors; - __pyx_t_13average_inner_uINT_t word_indices[0x2710]; - __pyx_t_13average_inner_uINT_t sent_adresses[0x2710]; - __pyx_t_13average_inner_uINT_t sentence_boundary[(0x2710 + 1)]; __pyx_t_13average_inner_uINT_t subwords_idx_len[0x2710]; __pyx_t_13average_inner_uINT_t *subwords_idx; }; @@ -1717,17 +1696,17 @@ static __pyx_t_13average_inner_REAL_t *__pyx_vp_13average_inner_ONEF = 0; #define __pyx_v_13average_inner_ONEF (*__pyx_vp_13average_inner_ONEF) static __pyx_t_13average_inner_REAL_t *__pyx_vp_13average_inner_ZEROF = 0; #define __pyx_v_13average_inner_ZEROF (*__pyx_vp_13average_inner_ZEROF) -static PyObject *(*__pyx_f_13average_inner_init_base_s2v_config)(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *, PyObject *); /*proto*/ -static PyObject *(*__pyx_f_13average_inner_init_ft_s2v_config)(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *, PyObject *); /*proto*/ -static PyObject *(*__pyx_f_13average_inner_populate_base_s2v_config)(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *); /*proto*/ -static PyObject *(*__pyx_f_13average_inner_populate_ft_s2v_config)(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *); /*proto*/ +static PyObject *(*__pyx_f_13average_inner_init_base_s2v_config)(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *, PyObject *); /*proto*/ +static PyObject *(*__pyx_f_13average_inner_init_ft_s2v_config)(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *, PyObject *); /*proto*/ +static PyObject *(*__pyx_f_13average_inner_populate_base_s2v_config)(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *); /*proto*/ +static PyObject *(*__pyx_f_13average_inner_populate_ft_s2v_config)(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *); /*proto*/ /* Module declarations from 'pooling_inner' */ static void __pyx_f_13pooling_inner_swrmax_pool(int const *, float const *, float const *, float *); /*proto*/ -static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ -static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_REAL_t); /*proto*/ -static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ -static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_REAL_t); /*proto*/ +static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t_13average_inner_VecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ +static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_VecsConfig *, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_REAL_t); /*proto*/ +static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_13average_inner_VecsConfig *, __pyx_t_13average_inner_uINT_t); /*proto*/ +static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __pyx_t_13average_inner_VecsConfig *, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_uINT_t, __pyx_t_13average_inner_REAL_t); /*proto*/ #define __Pyx_MODULE_NAME "pooling_inner" extern int __pyx_module_is_main_pooling_inner; int __pyx_module_is_main_pooling_inner = 0; @@ -1738,10 +1717,8 @@ static PyObject *__pyx_builtin_ValueError; static PyObject *__pyx_builtin_RuntimeError; static PyObject *__pyx_builtin_ImportError; static const char __pyx_k__8[] = "*"; -static const char __pyx_k_ft[] = "ft"; static const char __pyx_k_np[] = "np"; static const char __pyx_k_wv[] = "wv"; -static const char __pyx_k_w2v[] = "w2v"; static const char __pyx_k_init[] = "init"; static const char __pyx_k_main[] = "__main__"; static const char __pyx_k_name[] = "__name__"; @@ -1752,6 +1729,7 @@ static const char __pyx_k_model[] = "model"; static const char __pyx_k_numpy[] = "numpy"; static const char __pyx_k_range[] = "range"; static const char __pyx_k_vocab[] = "vocab"; +static const char __pyx_k_config[] = "config"; static const char __pyx_k_import[] = "__import__"; static const char __pyx_k_memory[] = "memory"; static const char __pyx_k_target[] = "target"; @@ -1792,10 +1770,10 @@ static PyObject *__pyx_n_s_RuntimeError; static PyObject *__pyx_n_s_ValueError; static PyObject *__pyx_n_s__8; static PyObject *__pyx_n_s_cline_in_traceback; +static PyObject *__pyx_n_s_config; static PyObject *__pyx_n_s_eff_sentences; static PyObject *__pyx_n_s_eff_words; static PyObject *__pyx_n_s_fblas; -static PyObject *__pyx_n_s_ft; static PyObject *__pyx_n_s_hierarchical; static PyObject *__pyx_n_s_import; static PyObject *__pyx_n_s_indexed_sentences; @@ -1820,7 +1798,6 @@ static PyObject *__pyx_n_s_test; static PyObject *__pyx_n_s_train_pooling_cy; static PyObject *__pyx_kp_u_unknown_dtype_code_in_numpy_pxd; static PyObject *__pyx_n_s_vocab; -static PyObject *__pyx_n_s_w2v; static PyObject *__pyx_n_s_window_size; static PyObject *__pyx_n_s_window_stride; static PyObject *__pyx_n_s_wv; @@ -1843,7 +1820,7 @@ static PyObject *__pyx_codeobj__10; static PyObject *__pyx_codeobj__11; /* Late includes */ -/* "pooling_inner.pyx":44 +/* "pooling_inner.pyx":43 * DEF MAX_NGRAMS = 40 * * cdef void swrmax_pool( # <<<<<<<<<<<<<< @@ -1856,7 +1833,7 @@ static void __pyx_f_13pooling_inner_swrmax_pool(int const *__pyx_v_N, float cons int __pyx_t_1; int __pyx_t_2; - /* "pooling_inner.pyx":65 + /* "pooling_inner.pyx":64 * """ * cdef int i * for i from 0 <= i < N[0] by 1: # <<<<<<<<<<<<<< @@ -1866,7 +1843,7 @@ static void __pyx_f_13pooling_inner_swrmax_pool(int const *__pyx_v_N, float cons __pyx_t_1 = (__pyx_v_N[0]); for (__pyx_v_i = 0; __pyx_v_i < __pyx_t_1; __pyx_v_i+=1) { - /* "pooling_inner.pyx":66 + /* "pooling_inner.pyx":65 * cdef int i * for i from 0 <= i < N[0] by 1: * if (alpha[0] * X[i]) > Y[i]: # <<<<<<<<<<<<<< @@ -1876,7 +1853,7 @@ static void __pyx_f_13pooling_inner_swrmax_pool(int const *__pyx_v_N, float cons __pyx_t_2 = ((((__pyx_v_alpha[0]) * (__pyx_v_X[__pyx_v_i])) > (__pyx_v_Y[__pyx_v_i])) != 0); if (__pyx_t_2) { - /* "pooling_inner.pyx":67 + /* "pooling_inner.pyx":66 * for i from 0 <= i < N[0] by 1: * if (alpha[0] * X[i]) > Y[i]: * Y[i] = alpha[0] * X[i] # <<<<<<<<<<<<<< @@ -1885,7 +1862,7 @@ static void __pyx_f_13pooling_inner_swrmax_pool(int const *__pyx_v_N, float cons */ (__pyx_v_Y[__pyx_v_i]) = ((__pyx_v_alpha[0]) * (__pyx_v_X[__pyx_v_i])); - /* "pooling_inner.pyx":66 + /* "pooling_inner.pyx":65 * cdef int i * for i from 0 <= i < N[0] by 1: * if (alpha[0] * X[i]) > Y[i]: # <<<<<<<<<<<<<< @@ -1895,7 +1872,7 @@ static void __pyx_f_13pooling_inner_swrmax_pool(int const *__pyx_v_N, float cons } } - /* "pooling_inner.pyx":44 + /* "pooling_inner.pyx":43 * DEF MAX_NGRAMS = 40 * * cdef void swrmax_pool( # <<<<<<<<<<<<<< @@ -1906,15 +1883,15 @@ static void __pyx_f_13pooling_inner_swrmax_pool(int const *__pyx_v_N, float cons /* function exit code */ } -/* "pooling_inner.pyx":69 +/* "pooling_inner.pyx":68 * Y[i] = alpha[0] * X[i] * * cdef void compute_base_sentence_pooling( # <<<<<<<<<<<<<< - * BaseSentenceVecsConfig *c, + * VecsConfig *c, * uINT_t num_sentences, */ -static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { +static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t_13average_inner_VecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { int __pyx_v_size; __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; @@ -1932,7 +1909,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t __pyx_t_13average_inner_uINT_t __pyx_t_6; __pyx_t_13average_inner_uINT_t __pyx_t_7; - /* "pooling_inner.pyx":88 + /* "pooling_inner.pyx":87 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -1942,7 +1919,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":96 + /* "pooling_inner.pyx":95 * REAL_t sent_len, inv_count * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -1954,7 +1931,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) { __pyx_v_sent_idx = __pyx_t_4; - /* "pooling_inner.pyx":97 + /* "pooling_inner.pyx":96 * * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -1963,7 +1940,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":98 + /* "pooling_inner.pyx":97 * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -1972,7 +1949,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":99 + /* "pooling_inner.pyx":98 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -1981,7 +1958,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":101 + /* "pooling_inner.pyx":100 * sent_len = ZEROF * * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -1993,7 +1970,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { __pyx_v_sent_pos = __pyx_t_7; - /* "pooling_inner.pyx":102 + /* "pooling_inner.pyx":101 * * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -2002,7 +1979,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":103 + /* "pooling_inner.pyx":102 * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< @@ -2011,7 +1988,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":104 + /* "pooling_inner.pyx":103 * sent_len += ONEF * sent_row = c.sent_adresses[sent_pos] * size * word_row = c.word_indices[sent_pos] * size # <<<<<<<<<<<<<< @@ -2020,7 +1997,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_sent_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":105 + /* "pooling_inner.pyx":104 * sent_row = c.sent_adresses[sent_pos] * size * word_row = c.word_indices[sent_pos] * size * word_idx = c.word_indices[sent_pos] # <<<<<<<<<<<<<< @@ -2029,7 +2006,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_sent_pos]); - /* "pooling_inner.pyx":107 + /* "pooling_inner.pyx":106 * word_idx = c.word_indices[sent_pos] * * swrmax_pool( # <<<<<<<<<<<<<< @@ -2040,26 +2017,26 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_pooling(struct __pyx_t } } - /* "pooling_inner.pyx":69 + /* "pooling_inner.pyx":68 * Y[i] = alpha[0] * X[i] * * cdef void compute_base_sentence_pooling( # <<<<<<<<<<<<<< - * BaseSentenceVecsConfig *c, + * VecsConfig *c, * uINT_t num_sentences, */ /* function exit code */ } -/* "pooling_inner.pyx":116 +/* "pooling_inner.pyx":115 * * * cdef void compute_base_sentence_hier_pooling( # <<<<<<<<<<<<<< - * BaseSentenceVecsConfig *c, + * VecsConfig *c, * uINT_t num_sentences, */ -static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences, __pyx_t_13average_inner_uINT_t __pyx_v_window_size, __pyx_t_13average_inner_REAL_t __pyx_v_window_stride) { +static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __pyx_t_13average_inner_VecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences, __pyx_t_13average_inner_uINT_t __pyx_v_window_size, __pyx_t_13average_inner_REAL_t __pyx_v_window_stride) { int __pyx_v_size; __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; @@ -2085,7 +2062,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ __pyx_t_13average_inner_uINT_t __pyx_t_10; __pyx_t_13average_inner_uINT_t __pyx_t_11; - /* "pooling_inner.pyx":141 + /* "pooling_inner.pyx":140 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -2095,7 +2072,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":149 + /* "pooling_inner.pyx":148 * REAL_t sent_len, win_len, inv_count * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -2107,7 +2084,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ for (__pyx_t_4 = 0; __pyx_t_4 < __pyx_t_3; __pyx_t_4+=1) { __pyx_v_sent_idx = __pyx_t_4; - /* "pooling_inner.pyx":150 + /* "pooling_inner.pyx":149 * * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -2116,7 +2093,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":151 + /* "pooling_inner.pyx":150 * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -2125,7 +2102,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":152 + /* "pooling_inner.pyx":151 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -2134,7 +2111,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":154 + /* "pooling_inner.pyx":153 * sent_len = ZEROF * * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -2146,7 +2123,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ for (__pyx_t_7 = __pyx_v_sent_start; __pyx_t_7 < __pyx_t_6; __pyx_t_7+=1) { __pyx_v_sent_pos = __pyx_t_7; - /* "pooling_inner.pyx":155 + /* "pooling_inner.pyx":154 * * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -2155,7 +2132,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":157 + /* "pooling_inner.pyx":156 * sent_len += ONEF * * if (sent_len-ONEF) % window_stride != ZEROF: # <<<<<<<<<<<<<< @@ -2165,7 +2142,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ __pyx_t_8 = ((fmodf((__pyx_v_sent_len - __pyx_v_13average_inner_ONEF), __pyx_v_window_stride) != __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_8) { - /* "pooling_inner.pyx":158 + /* "pooling_inner.pyx":157 * * if (sent_len-ONEF) % window_stride != ZEROF: * continue # <<<<<<<<<<<<<< @@ -2174,7 +2151,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ goto __pyx_L5_continue; - /* "pooling_inner.pyx":157 + /* "pooling_inner.pyx":156 * sent_len += ONEF * * if (sent_len-ONEF) % window_stride != ZEROF: # <<<<<<<<<<<<<< @@ -2183,7 +2160,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ } - /* "pooling_inner.pyx":160 + /* "pooling_inner.pyx":159 * continue * * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< @@ -2192,7 +2169,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":162 + /* "pooling_inner.pyx":161 * sent_row = c.sent_adresses[sent_pos] * size * * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< @@ -2202,7 +2179,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ __pyx_t_8 = (((__pyx_v_sent_pos + __pyx_v_window_size) > __pyx_v_sent_end) != 0); if (__pyx_t_8) { - /* "pooling_inner.pyx":163 + /* "pooling_inner.pyx":162 * * if sent_pos + window_size > sent_end: * window_end = sent_end # <<<<<<<<<<<<<< @@ -2211,7 +2188,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_window_end = __pyx_v_sent_end; - /* "pooling_inner.pyx":162 + /* "pooling_inner.pyx":161 * sent_row = c.sent_adresses[sent_pos] * size * * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< @@ -2221,7 +2198,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ goto __pyx_L8; } - /* "pooling_inner.pyx":165 + /* "pooling_inner.pyx":164 * window_end = sent_end * else: * window_end = sent_pos + window_size # <<<<<<<<<<<<<< @@ -2233,7 +2210,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ } __pyx_L8:; - /* "pooling_inner.pyx":168 + /* "pooling_inner.pyx":167 * * # Compute the locally averaged window * win_len = ZEROF # <<<<<<<<<<<<<< @@ -2242,7 +2219,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_win_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":169 + /* "pooling_inner.pyx":168 * # Compute the locally averaged window * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2251,7 +2228,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":170 + /* "pooling_inner.pyx":169 * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2260,7 +2237,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":171 + /* "pooling_inner.pyx":170 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * for window_pos in range(sent_pos, window_end): # <<<<<<<<<<<<<< @@ -2272,7 +2249,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ for (__pyx_t_11 = __pyx_v_sent_pos; __pyx_t_11 < __pyx_t_10; __pyx_t_11+=1) { __pyx_v_window_pos = __pyx_t_11; - /* "pooling_inner.pyx":172 + /* "pooling_inner.pyx":171 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * for window_pos in range(sent_pos, window_end): * win_len += ONEF # <<<<<<<<<<<<<< @@ -2281,7 +2258,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_win_len = (__pyx_v_win_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":174 + /* "pooling_inner.pyx":173 * win_len += ONEF * * word_row = c.word_indices[window_pos] * size # <<<<<<<<<<<<<< @@ -2290,7 +2267,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_window_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":175 + /* "pooling_inner.pyx":174 * * word_row = c.word_indices[window_pos] * size * word_idx = c.word_indices[window_pos] # <<<<<<<<<<<<<< @@ -2299,7 +2276,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_window_pos]); - /* "pooling_inner.pyx":177 + /* "pooling_inner.pyx":176 * word_idx = c.word_indices[window_pos] * * saxpy( # <<<<<<<<<<<<<< @@ -2309,7 +2286,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); } - /* "pooling_inner.pyx":187 + /* "pooling_inner.pyx":186 * * # Rescale for dynamic window size * if win_len > ZEROF: # <<<<<<<<<<<<<< @@ -2319,7 +2296,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ __pyx_t_8 = ((__pyx_v_win_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_8) { - /* "pooling_inner.pyx":188 + /* "pooling_inner.pyx":187 * # Rescale for dynamic window size * if win_len > ZEROF: * inv_count = ONEF / win_len # <<<<<<<<<<<<<< @@ -2328,7 +2305,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_win_len); - /* "pooling_inner.pyx":189 + /* "pooling_inner.pyx":188 * if win_len > ZEROF: * inv_count = ONEF / win_len * saxpy( # <<<<<<<<<<<<<< @@ -2337,7 +2314,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE)); - /* "pooling_inner.pyx":187 + /* "pooling_inner.pyx":186 * * # Rescale for dynamic window size * if win_len > ZEROF: # <<<<<<<<<<<<<< @@ -2346,7 +2323,7 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ */ } - /* "pooling_inner.pyx":198 + /* "pooling_inner.pyx":197 * ) * * swrmax_pool( # <<<<<<<<<<<<<< @@ -2358,26 +2335,26 @@ static void __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling(struct __ } } - /* "pooling_inner.pyx":116 + /* "pooling_inner.pyx":115 * * * cdef void compute_base_sentence_hier_pooling( # <<<<<<<<<<<<<< - * BaseSentenceVecsConfig *c, + * VecsConfig *c, * uINT_t num_sentences, */ /* function exit code */ } -/* "pooling_inner.pyx":205 +/* "pooling_inner.pyx":204 * ) * * cdef void compute_ft_sentence_pooling( # <<<<<<<<<<<<<< - * FTSentenceVecsConfig *c, + * VecsConfig *c, * uINT_t num_sentences, */ -static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { +static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_13average_inner_VecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences) { int __pyx_v_size; __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; @@ -2405,7 +2382,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_13average_inner_uINT_t __pyx_t_11; __pyx_t_13average_inner_uINT_t __pyx_t_12; - /* "pooling_inner.pyx":224 + /* "pooling_inner.pyx":223 * """ * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -2415,7 +2392,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":234 + /* "pooling_inner.pyx":233 * REAL_t sent_len * REAL_t inv_count, inv_ngram * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< @@ -2425,7 +2402,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_2 = __pyx_v_c->oov_weight; __pyx_v_oov_weight = __pyx_t_2; - /* "pooling_inner.pyx":236 + /* "pooling_inner.pyx":235 * REAL_t oov_weight = c.oov_weight * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -2437,7 +2414,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { __pyx_v_sent_idx = __pyx_t_5; - /* "pooling_inner.pyx":237 + /* "pooling_inner.pyx":236 * * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -2446,7 +2423,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":238 + /* "pooling_inner.pyx":237 * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -2455,7 +2432,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":239 + /* "pooling_inner.pyx":238 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -2464,7 +2441,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":241 + /* "pooling_inner.pyx":240 * sent_len = ZEROF * * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -2476,7 +2453,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { __pyx_v_sent_pos = __pyx_t_8; - /* "pooling_inner.pyx":242 + /* "pooling_inner.pyx":241 * * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -2485,7 +2462,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":243 + /* "pooling_inner.pyx":242 * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< @@ -2494,7 +2471,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":245 + /* "pooling_inner.pyx":244 * sent_row = c.sent_adresses[sent_pos] * size * * word_idx = c.word_indices[sent_pos] # <<<<<<<<<<<<<< @@ -2503,7 +2480,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_sent_pos]); - /* "pooling_inner.pyx":246 + /* "pooling_inner.pyx":245 * * word_idx = c.word_indices[sent_pos] * ngrams = c.subwords_idx_len[sent_pos] # <<<<<<<<<<<<<< @@ -2512,7 +2489,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_sent_pos]); - /* "pooling_inner.pyx":248 + /* "pooling_inner.pyx":247 * ngrams = c.subwords_idx_len[sent_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2522,7 +2499,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":249 + /* "pooling_inner.pyx":248 * * if ngrams == 0: * word_row = c.word_indices[sent_pos] * size # <<<<<<<<<<<<<< @@ -2531,7 +2508,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_sent_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":251 + /* "pooling_inner.pyx":250 * word_row = c.word_indices[sent_pos] * size * * swrmax_pool( # <<<<<<<<<<<<<< @@ -2540,7 +2517,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_f_13pooling_inner_swrmax_pool((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&(__pyx_v_c->sentence_vectors[__pyx_v_sent_row]))); - /* "pooling_inner.pyx":248 + /* "pooling_inner.pyx":247 * ngrams = c.subwords_idx_len[sent_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2550,7 +2527,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 goto __pyx_L7; } - /* "pooling_inner.pyx":259 + /* "pooling_inner.pyx":258 * * else: * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2560,7 +2537,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 /*else*/ { (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":260 + /* "pooling_inner.pyx":259 * else: * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< @@ -2569,7 +2546,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); - /* "pooling_inner.pyx":261 + /* "pooling_inner.pyx":260 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): # <<<<<<<<<<<<<< @@ -2581,7 +2558,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 for (__pyx_t_12 = 0; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { __pyx_v_ngram_pos = __pyx_t_12; - /* "pooling_inner.pyx":262 + /* "pooling_inner.pyx":261 * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size # <<<<<<<<<<<<<< @@ -2590,7 +2567,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 */ __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_sent_pos * 40) + __pyx_v_ngram_pos)]) * __pyx_v_size); - /* "pooling_inner.pyx":263 + /* "pooling_inner.pyx":262 * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(sent_pos * MAX_NGRAMS)+ngram_pos] * size * saxpy( # <<<<<<<<<<<<<< @@ -2600,7 +2577,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_ngram), (&(__pyx_v_c->ngram_vectors[__pyx_v_ngram_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); } - /* "pooling_inner.pyx":272 + /* "pooling_inner.pyx":271 * ) * * swrmax_pool( # <<<<<<<<<<<<<< @@ -2613,26 +2590,26 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_pooling(struct __pyx_t_1 } } - /* "pooling_inner.pyx":205 + /* "pooling_inner.pyx":204 * ) * * cdef void compute_ft_sentence_pooling( # <<<<<<<<<<<<<< - * FTSentenceVecsConfig *c, + * VecsConfig *c, * uINT_t num_sentences, */ /* function exit code */ } -/* "pooling_inner.pyx":280 +/* "pooling_inner.pyx":279 * # There's nothing to do here for many-to-one mappings * * cdef void compute_ft_sentence_hier_pooling( # <<<<<<<<<<<<<< - * FTSentenceVecsConfig *c, + * VecsConfig *c, * uINT_t num_sentences, */ -static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __pyx_t_13average_inner_FTSentenceVecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences, __pyx_t_13average_inner_uINT_t __pyx_v_window_size, __pyx_t_13average_inner_REAL_t __pyx_v_window_stride) { +static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __pyx_t_13average_inner_VecsConfig *__pyx_v_c, __pyx_t_13average_inner_uINT_t __pyx_v_num_sentences, __pyx_t_13average_inner_uINT_t __pyx_v_window_size, __pyx_t_13average_inner_REAL_t __pyx_v_window_stride) { int __pyx_v_size; __pyx_t_13average_inner_uINT_t __pyx_v_sent_idx; __pyx_t_13average_inner_uINT_t __pyx_v_sent_start; @@ -2667,7 +2644,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_13average_inner_uINT_t __pyx_t_14; __pyx_t_13average_inner_uINT_t __pyx_t_15; - /* "pooling_inner.pyx":306 + /* "pooling_inner.pyx":305 * * cdef: * int size = c.size # <<<<<<<<<<<<<< @@ -2677,7 +2654,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_1 = __pyx_v_c->size; __pyx_v_size = __pyx_t_1; - /* "pooling_inner.pyx":316 + /* "pooling_inner.pyx":315 * REAL_t sent_len, win_len * REAL_t inv_count, inv_ngram * REAL_t oov_weight = c.oov_weight # <<<<<<<<<<<<<< @@ -2687,7 +2664,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_2 = __pyx_v_c->oov_weight; __pyx_v_oov_weight = __pyx_t_2; - /* "pooling_inner.pyx":318 + /* "pooling_inner.pyx":317 * REAL_t oov_weight = c.oov_weight * * for sent_idx in range(num_sentences): # <<<<<<<<<<<<<< @@ -2699,7 +2676,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_5 = 0; __pyx_t_5 < __pyx_t_4; __pyx_t_5+=1) { __pyx_v_sent_idx = __pyx_t_5; - /* "pooling_inner.pyx":319 + /* "pooling_inner.pyx":318 * * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] # <<<<<<<<<<<<<< @@ -2708,7 +2685,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_start = (__pyx_v_c->sentence_boundary[__pyx_v_sent_idx]); - /* "pooling_inner.pyx":320 + /* "pooling_inner.pyx":319 * for sent_idx in range(num_sentences): * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] # <<<<<<<<<<<<<< @@ -2717,7 +2694,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_end = (__pyx_v_c->sentence_boundary[(__pyx_v_sent_idx + 1)]); - /* "pooling_inner.pyx":321 + /* "pooling_inner.pyx":320 * sent_start = c.sentence_boundary[sent_idx] * sent_end = c.sentence_boundary[sent_idx + 1] * sent_len = ZEROF # <<<<<<<<<<<<<< @@ -2726,7 +2703,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":323 + /* "pooling_inner.pyx":322 * sent_len = ZEROF * * for sent_pos in range(sent_start, sent_end): # <<<<<<<<<<<<<< @@ -2738,7 +2715,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_8 = __pyx_v_sent_start; __pyx_t_8 < __pyx_t_7; __pyx_t_8+=1) { __pyx_v_sent_pos = __pyx_t_8; - /* "pooling_inner.pyx":324 + /* "pooling_inner.pyx":323 * * for sent_pos in range(sent_start, sent_end): * sent_len += ONEF # <<<<<<<<<<<<<< @@ -2747,7 +2724,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_len = (__pyx_v_sent_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":326 + /* "pooling_inner.pyx":325 * sent_len += ONEF * * if (sent_len-ONEF) % window_stride != ZEROF: # <<<<<<<<<<<<<< @@ -2757,7 +2734,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = ((fmodf((__pyx_v_sent_len - __pyx_v_13average_inner_ONEF), __pyx_v_window_stride) != __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":327 + /* "pooling_inner.pyx":326 * * if (sent_len-ONEF) % window_stride != ZEROF: * continue # <<<<<<<<<<<<<< @@ -2766,7 +2743,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ goto __pyx_L5_continue; - /* "pooling_inner.pyx":326 + /* "pooling_inner.pyx":325 * sent_len += ONEF * * if (sent_len-ONEF) % window_stride != ZEROF: # <<<<<<<<<<<<<< @@ -2775,7 +2752,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ } - /* "pooling_inner.pyx":329 + /* "pooling_inner.pyx":328 * continue * * sent_row = c.sent_adresses[sent_pos] * size # <<<<<<<<<<<<<< @@ -2784,7 +2761,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_sent_row = ((__pyx_v_c->sent_adresses[__pyx_v_sent_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":331 + /* "pooling_inner.pyx":330 * sent_row = c.sent_adresses[sent_pos] * size * * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< @@ -2794,7 +2771,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = (((__pyx_v_sent_pos + __pyx_v_window_size) > __pyx_v_sent_end) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":332 + /* "pooling_inner.pyx":331 * * if sent_pos + window_size > sent_end: * window_end = sent_end # <<<<<<<<<<<<<< @@ -2803,7 +2780,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_window_end = __pyx_v_sent_end; - /* "pooling_inner.pyx":331 + /* "pooling_inner.pyx":330 * sent_row = c.sent_adresses[sent_pos] * size * * if sent_pos + window_size > sent_end: # <<<<<<<<<<<<<< @@ -2813,7 +2790,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py goto __pyx_L8; } - /* "pooling_inner.pyx":334 + /* "pooling_inner.pyx":333 * window_end = sent_end * else: * window_end = sent_pos + window_size # <<<<<<<<<<<<<< @@ -2825,7 +2802,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py } __pyx_L8:; - /* "pooling_inner.pyx":337 + /* "pooling_inner.pyx":336 * * # Compute the locally averaged window * win_len = ZEROF # <<<<<<<<<<<<<< @@ -2834,7 +2811,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_win_len = __pyx_v_13average_inner_ZEROF; - /* "pooling_inner.pyx":338 + /* "pooling_inner.pyx":337 * # Compute the locally averaged window * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2843,7 +2820,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ (void)(memset(__pyx_v_c->mem, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":339 + /* "pooling_inner.pyx":338 * win_len = ZEROF * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2852,7 +2829,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":340 + /* "pooling_inner.pyx":339 * memset(c.mem, 0, size * cython.sizeof(REAL_t)) * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * for window_pos in range(sent_pos, window_end): # <<<<<<<<<<<<<< @@ -2864,7 +2841,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_12 = __pyx_v_sent_pos; __pyx_t_12 < __pyx_t_11; __pyx_t_12+=1) { __pyx_v_window_pos = __pyx_t_12; - /* "pooling_inner.pyx":341 + /* "pooling_inner.pyx":340 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * for window_pos in range(sent_pos, window_end): * win_len += ONEF # <<<<<<<<<<<<<< @@ -2873,7 +2850,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_win_len = (__pyx_v_win_len + __pyx_v_13average_inner_ONEF); - /* "pooling_inner.pyx":342 + /* "pooling_inner.pyx":341 * for window_pos in range(sent_pos, window_end): * win_len += ONEF * ngrams = c.subwords_idx_len[window_pos] # <<<<<<<<<<<<<< @@ -2882,7 +2859,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_ngrams = (__pyx_v_c->subwords_idx_len[__pyx_v_window_pos]); - /* "pooling_inner.pyx":344 + /* "pooling_inner.pyx":343 * ngrams = c.subwords_idx_len[window_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2892,7 +2869,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = ((__pyx_v_ngrams == 0) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":345 + /* "pooling_inner.pyx":344 * * if ngrams == 0: * word_row = c.word_indices[window_pos] * size # <<<<<<<<<<<<<< @@ -2901,7 +2878,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_word_row = ((__pyx_v_c->word_indices[__pyx_v_window_pos]) * __pyx_v_size); - /* "pooling_inner.pyx":346 + /* "pooling_inner.pyx":345 * if ngrams == 0: * word_row = c.word_indices[window_pos] * size * word_idx = c.word_indices[window_pos] # <<<<<<<<<<<<<< @@ -2910,7 +2887,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_word_idx = (__pyx_v_c->word_indices[__pyx_v_window_pos]); - /* "pooling_inner.pyx":348 + /* "pooling_inner.pyx":347 * word_idx = c.word_indices[window_pos] * * saxpy( # <<<<<<<<<<<<<< @@ -2919,7 +2896,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&(__pyx_v_c->word_weights[__pyx_v_word_idx])), (&(__pyx_v_c->word_vectors[__pyx_v_word_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE)); - /* "pooling_inner.pyx":344 + /* "pooling_inner.pyx":343 * ngrams = c.subwords_idx_len[window_pos] * * if ngrams == 0: # <<<<<<<<<<<<<< @@ -2929,7 +2906,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py goto __pyx_L11; } - /* "pooling_inner.pyx":358 + /* "pooling_inner.pyx":357 * * else: * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -2939,7 +2916,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py /*else*/ { (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":359 + /* "pooling_inner.pyx":358 * else: * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight # <<<<<<<<<<<<<< @@ -2948,7 +2925,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_inv_ngram = ((__pyx_v_13average_inner_ONEF / ((__pyx_t_13average_inner_REAL_t)__pyx_v_ngrams)) * __pyx_v_c->oov_weight); - /* "pooling_inner.pyx":360 + /* "pooling_inner.pyx":359 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): # <<<<<<<<<<<<<< @@ -2960,7 +2937,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py for (__pyx_t_15 = 0; __pyx_t_15 < __pyx_t_14; __pyx_t_15+=1) { __pyx_v_ngram_pos = __pyx_t_15; - /* "pooling_inner.pyx":361 + /* "pooling_inner.pyx":360 * inv_ngram = (ONEF / ngrams) * c.oov_weight * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(window_pos * MAX_NGRAMS)+ngram_pos] * size # <<<<<<<<<<<<<< @@ -2969,7 +2946,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_ngram_row = ((__pyx_v_c->subwords_idx[((__pyx_v_window_pos * 40) + __pyx_v_ngram_pos)]) * __pyx_v_size); - /* "pooling_inner.pyx":362 + /* "pooling_inner.pyx":361 * for ngram_pos in range(ngrams): * ngram_row = c.subwords_idx[(window_pos * MAX_NGRAMS)+ngram_pos] * size * saxpy( # <<<<<<<<<<<<<< @@ -2979,7 +2956,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_ngram), (&(__pyx_v_c->ngram_vectors[__pyx_v_ngram_row])), (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE)); } - /* "pooling_inner.pyx":371 + /* "pooling_inner.pyx":370 * ) * * saxpy( # <<<<<<<<<<<<<< @@ -2991,7 +2968,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_L11:; } - /* "pooling_inner.pyx":380 + /* "pooling_inner.pyx":379 * ) * * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) # <<<<<<<<<<<<<< @@ -3000,7 +2977,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ (void)(memset(__pyx_v_c->mem2, 0, (__pyx_v_size * (sizeof(__pyx_t_13average_inner_REAL_t))))); - /* "pooling_inner.pyx":382 + /* "pooling_inner.pyx":381 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * # Rescale for dynamic window size * if win_len > ZEROF: # <<<<<<<<<<<<<< @@ -3010,7 +2987,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py __pyx_t_9 = ((__pyx_v_win_len > __pyx_v_13average_inner_ZEROF) != 0); if (__pyx_t_9) { - /* "pooling_inner.pyx":383 + /* "pooling_inner.pyx":382 * # Rescale for dynamic window size * if win_len > ZEROF: * inv_count = ONEF / win_len # <<<<<<<<<<<<<< @@ -3019,7 +2996,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_inv_count = (__pyx_v_13average_inner_ONEF / __pyx_v_win_len); - /* "pooling_inner.pyx":384 + /* "pooling_inner.pyx":383 * if win_len > ZEROF: * inv_count = ONEF / win_len * saxpy( # <<<<<<<<<<<<<< @@ -3028,7 +3005,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ __pyx_v_13average_inner_saxpy((&__pyx_v_size), (&__pyx_v_inv_count), __pyx_v_c->mem, (&__pyx_v_13average_inner_ONE), __pyx_v_c->mem2, (&__pyx_v_13average_inner_ONE)); - /* "pooling_inner.pyx":382 + /* "pooling_inner.pyx":381 * memset(c.mem2, 0, size * cython.sizeof(REAL_t)) * # Rescale for dynamic window size * if win_len > ZEROF: # <<<<<<<<<<<<<< @@ -3037,7 +3014,7 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py */ } - /* "pooling_inner.pyx":393 + /* "pooling_inner.pyx":392 * ) * * swrmax_pool( # <<<<<<<<<<<<<< @@ -3049,18 +3026,18 @@ static void __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling(struct __py } } - /* "pooling_inner.pyx":280 + /* "pooling_inner.pyx":279 * # There's nothing to do here for many-to-one mappings * * cdef void compute_ft_sentence_hier_pooling( # <<<<<<<<<<<<<< - * FTSentenceVecsConfig *c, + * VecsConfig *c, * uINT_t num_sentences, */ /* function exit code */ } -/* "pooling_inner.pyx":401 +/* "pooling_inner.pyx":400 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< @@ -3107,23 +3084,23 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self case 1: if (likely((values[1] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_indexed_sentences)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 401, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 1); __PYX_ERR(0, 400, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 2: if (likely((values[2] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_target)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 401, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 2); __PYX_ERR(0, 400, __pyx_L3_error) } CYTHON_FALLTHROUGH; case 3: if (likely((values[3] = __Pyx_PyDict_GetItemStr(__pyx_kwds, __pyx_n_s_memory)) != 0)) kw_args--; else { - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 401, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, 3); __PYX_ERR(0, 400, __pyx_L3_error) } } if (unlikely(kw_args > 0)) { - if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 401, __pyx_L3_error) + if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, pos_args, "train_pooling_cy") < 0)) __PYX_ERR(0, 400, __pyx_L3_error) } } else if (PyTuple_GET_SIZE(__pyx_args) != 4) { goto __pyx_L5_argtuple_error; @@ -3140,7 +3117,7 @@ static PyObject *__pyx_pw_13pooling_inner_1train_pooling_cy(PyObject *__pyx_self } goto __pyx_L4_argument_unpacking_done; __pyx_L5_argtuple_error:; - __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 401, __pyx_L3_error) + __Pyx_RaiseArgtupleInvalid("train_pooling_cy", 1, 4, 4, PyTuple_GET_SIZE(__pyx_args)); __PYX_ERR(0, 400, __pyx_L3_error) __pyx_L3_error:; __Pyx_AddTraceback("pooling_inner.train_pooling_cy", __pyx_clineno, __pyx_lineno, __pyx_filename); __Pyx_RefNannyFinishContext(); @@ -3158,8 +3135,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_13average_inner_uINT_t __pyx_v_eff_words; __pyx_t_13average_inner_uINT_t __pyx_v_window_size; __pyx_t_13average_inner_REAL_t __pyx_v_window_stride; - struct __pyx_t_13average_inner_BaseSentenceVecsConfig __pyx_v_w2v; - struct __pyx_t_13average_inner_FTSentenceVecsConfig __pyx_v_ft; + struct __pyx_t_13average_inner_VecsConfig __pyx_v_config; PyObject *__pyx_r = NULL; __Pyx_RefNannyDeclarations PyObject *__pyx_t_1 = NULL; @@ -3174,7 +3150,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_13average_inner_uINT_t __pyx_t_10; __Pyx_RefNannySetupContext("train_pooling_cy", 0); - /* "pooling_inner.pyx":430 + /* "pooling_inner.pyx":429 * """ * * cdef uINT_t eff_sentences = 0 # <<<<<<<<<<<<<< @@ -3183,7 +3159,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_sentences = 0; - /* "pooling_inner.pyx":431 + /* "pooling_inner.pyx":430 * * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 # <<<<<<<<<<<<<< @@ -3192,78 +3168,78 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec */ __pyx_v_eff_words = 0; - /* "pooling_inner.pyx":432 + /* "pooling_inner.pyx":431 * cdef uINT_t eff_sentences = 0 * cdef uINT_t eff_words = 0 * cdef uINT_t window_size = model.window_size # <<<<<<<<<<<<<< * cdef REAL_t window_stride = model.window_stride - * cdef BaseSentenceVecsConfig w2v + * cdef VecsConfig config */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_size); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 432, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_size); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 431, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_1); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 432, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_1); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 431, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_v_window_size = ((__pyx_t_13average_inner_uINT_t)__pyx_t_2); - /* "pooling_inner.pyx":433 + /* "pooling_inner.pyx":432 * cdef uINT_t eff_words = 0 * cdef uINT_t window_size = model.window_size * cdef REAL_t window_stride = model.window_stride # <<<<<<<<<<<<<< - * cdef BaseSentenceVecsConfig w2v - * cdef FTSentenceVecsConfig ft + * cdef VecsConfig config + * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_stride); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 433, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_window_stride); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 432, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_3 = __pyx_PyFloat_AsFloat(__pyx_t_1); if (unlikely((__pyx_t_3 == ((npy_float32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 433, __pyx_L1_error) + __pyx_t_3 = __pyx_PyFloat_AsFloat(__pyx_t_1); if (unlikely((__pyx_t_3 == ((npy_float32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 432, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_v_window_stride = ((__pyx_t_13average_inner_REAL_t)__pyx_t_3); - /* "pooling_inner.pyx":437 - * cdef FTSentenceVecsConfig ft + /* "pooling_inner.pyx":435 + * cdef VecsConfig config * * if not model.is_ft: # <<<<<<<<<<<<<< - * init_base_s2v_config(&w2v, model, target, memory) + * init_base_s2v_config(&config, model, target, memory) * */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 437, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_is_ft); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 435, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 437, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 435, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_5 = ((!__pyx_t_4) != 0); if (__pyx_t_5) { - /* "pooling_inner.pyx":438 + /* "pooling_inner.pyx":436 * * if not model.is_ft: - * init_base_s2v_config(&w2v, model, target, memory) # <<<<<<<<<<<<<< + * init_base_s2v_config(&config, model, target, memory) # <<<<<<<<<<<<<< * * eff_sentences, eff_words = populate_base_s2v_config( */ - __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_w2v), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 438, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_base_s2v_config((&__pyx_v_config), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 436, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":442 + /* "pooling_inner.pyx":440 * eff_sentences, eff_words = populate_base_s2v_config( - * &w2v, + * &config, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 442, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 440, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_6 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 442, __pyx_L1_error) + __pyx_t_6 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 440, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":440 - * init_base_s2v_config(&w2v, model, target, memory) + /* "pooling_inner.pyx":438 + * init_base_s2v_config(&config, model, target, memory) * * eff_sentences, eff_words = populate_base_s2v_config( # <<<<<<<<<<<<<< - * &w2v, + * &config, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_w2v), __pyx_t_6, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 440, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_base_s2v_config((&__pyx_v_config), __pyx_t_6, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 438, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -3272,7 +3248,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 440, __pyx_L1_error) + __PYX_ERR(0, 438, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -3285,15 +3261,15 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_6); __Pyx_INCREF(__pyx_t_7); #else - __pyx_t_6 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 440, __pyx_L1_error) + __pyx_t_6 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 438, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); - __pyx_t_7 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 440, __pyx_L1_error) + __pyx_t_7 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 438, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_8 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_8)) __PYX_ERR(0, 440, __pyx_L1_error) + __pyx_t_8 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_8)) __PYX_ERR(0, 438, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_8); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_9 = Py_TYPE(__pyx_t_8)->tp_iternext; @@ -3301,7 +3277,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_6); index = 1; __pyx_t_7 = __pyx_t_9(__pyx_t_8); if (unlikely(!__pyx_t_7)) goto __pyx_L4_unpacking_failed; __Pyx_GOTREF(__pyx_t_7); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_9(__pyx_t_8), 2) < 0) __PYX_ERR(0, 440, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_9(__pyx_t_8), 2) < 0) __PYX_ERR(0, 438, __pyx_L1_error) __pyx_t_9 = NULL; __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; goto __pyx_L5_unpacking_done; @@ -3309,36 +3285,36 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; __pyx_t_9 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 440, __pyx_L1_error) + __PYX_ERR(0, 438, __pyx_L1_error) __pyx_L5_unpacking_done:; } - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 440, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 438, __pyx_L1_error) __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; - __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_7); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 440, __pyx_L1_error) + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_7); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 438, __pyx_L1_error) __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; __pyx_v_eff_sentences = __pyx_t_2; __pyx_v_eff_words = __pyx_t_10; - /* "pooling_inner.pyx":446 + /* "pooling_inner.pyx":444 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: * compute_base_sentence_pooling( */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 446, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 444, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 446, __pyx_L1_error) + __pyx_t_5 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_5 < 0)) __PYX_ERR(0, 444, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_4 = ((!__pyx_t_5) != 0); if (__pyx_t_4) { - /* "pooling_inner.pyx":447 + /* "pooling_inner.pyx":445 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< * compute_base_sentence_pooling( - * &w2v, + * &config, */ { #ifdef WITH_THREAD @@ -3348,22 +3324,22 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":448 + /* "pooling_inner.pyx":446 * if not model.hierarchical: * with nogil: * compute_base_sentence_pooling( # <<<<<<<<<<<<<< - * &w2v, + * &config, * eff_sentences */ - __pyx_f_13pooling_inner_compute_base_sentence_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences); + __pyx_f_13pooling_inner_compute_base_sentence_pooling((&__pyx_v_config), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":447 + /* "pooling_inner.pyx":445 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< * compute_base_sentence_pooling( - * &w2v, + * &config, */ /*finally:*/ { /*normal exit:*/{ @@ -3377,7 +3353,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } } - /* "pooling_inner.pyx":446 + /* "pooling_inner.pyx":444 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< @@ -3387,12 +3363,12 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L6; } - /* "pooling_inner.pyx":453 + /* "pooling_inner.pyx":451 * ) * else: * with nogil: # <<<<<<<<<<<<<< * compute_base_sentence_hier_pooling( - * &w2v, + * &config, */ /*else*/ { { @@ -3403,22 +3379,22 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":454 + /* "pooling_inner.pyx":452 * else: * with nogil: * compute_base_sentence_hier_pooling( # <<<<<<<<<<<<<< - * &w2v, + * &config, * eff_sentences, */ - __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling((&__pyx_v_w2v), __pyx_v_eff_sentences, __pyx_v_window_size, __pyx_v_window_stride); + __pyx_f_13pooling_inner_compute_base_sentence_hier_pooling((&__pyx_v_config), __pyx_v_eff_sentences, __pyx_v_window_size, __pyx_v_window_stride); } - /* "pooling_inner.pyx":453 + /* "pooling_inner.pyx":451 * ) * else: * with nogil: # <<<<<<<<<<<<<< * compute_base_sentence_hier_pooling( - * &w2v, + * &config, */ /*finally:*/ { /*normal exit:*/{ @@ -3434,49 +3410,49 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } __pyx_L6:; - /* "pooling_inner.pyx":437 - * cdef FTSentenceVecsConfig ft + /* "pooling_inner.pyx":435 + * cdef VecsConfig config * * if not model.is_ft: # <<<<<<<<<<<<<< - * init_base_s2v_config(&w2v, model, target, memory) + * init_base_s2v_config(&config, model, target, memory) * */ goto __pyx_L3; } - /* "pooling_inner.pyx":461 + /* "pooling_inner.pyx":459 * ) * else: - * init_ft_s2v_config(&ft, model, target, memory) # <<<<<<<<<<<<<< + * init_ft_s2v_config(&config, model, target, memory) # <<<<<<<<<<<<<< * * eff_sentences, eff_words = populate_ft_s2v_config( */ /*else*/ { - __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_ft), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 461, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_init_ft_s2v_config((&__pyx_v_config), __pyx_v_model, __pyx_v_target, __pyx_v_memory); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 459, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":465 + /* "pooling_inner.pyx":463 * eff_sentences, eff_words = populate_ft_s2v_config( - * &ft, + * &config, * model.wv.vocab, # <<<<<<<<<<<<<< * indexed_sentences * ) */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 465, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_wv); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 463, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_7 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 465, __pyx_L1_error) + __pyx_t_7 = __Pyx_PyObject_GetAttrStr(__pyx_t_1, __pyx_n_s_vocab); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 463, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; - /* "pooling_inner.pyx":463 - * init_ft_s2v_config(&ft, model, target, memory) + /* "pooling_inner.pyx":461 + * init_ft_s2v_config(&config, model, target, memory) * * eff_sentences, eff_words = populate_ft_s2v_config( # <<<<<<<<<<<<<< - * &ft, + * &config, * model.wv.vocab, */ - __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_ft), __pyx_t_7, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 463, __pyx_L1_error) + __pyx_t_1 = __pyx_f_13average_inner_populate_ft_s2v_config((&__pyx_v_config), __pyx_t_7, __pyx_v_indexed_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 461, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) { @@ -3485,7 +3461,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec if (unlikely(size != 2)) { if (size > 2) __Pyx_RaiseTooManyValuesError(2); else if (size >= 0) __Pyx_RaiseNeedMoreValuesError(size); - __PYX_ERR(0, 463, __pyx_L1_error) + __PYX_ERR(0, 461, __pyx_L1_error) } #if CYTHON_ASSUME_SAFE_MACROS && !CYTHON_AVOID_BORROWED_REFS if (likely(PyTuple_CheckExact(sequence))) { @@ -3498,15 +3474,15 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_INCREF(__pyx_t_7); __Pyx_INCREF(__pyx_t_6); #else - __pyx_t_7 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 463, __pyx_L1_error) + __pyx_t_7 = PySequence_ITEM(sequence, 0); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 461, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); - __pyx_t_6 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 463, __pyx_L1_error) + __pyx_t_6 = PySequence_ITEM(sequence, 1); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 461, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); #endif __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; } else { Py_ssize_t index = -1; - __pyx_t_8 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_8)) __PYX_ERR(0, 463, __pyx_L1_error) + __pyx_t_8 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_8)) __PYX_ERR(0, 461, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_8); __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_9 = Py_TYPE(__pyx_t_8)->tp_iternext; @@ -3514,7 +3490,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_GOTREF(__pyx_t_7); index = 1; __pyx_t_6 = __pyx_t_9(__pyx_t_8); if (unlikely(!__pyx_t_6)) goto __pyx_L13_unpacking_failed; __Pyx_GOTREF(__pyx_t_6); - if (__Pyx_IternextUnpackEndCheck(__pyx_t_9(__pyx_t_8), 2) < 0) __PYX_ERR(0, 463, __pyx_L1_error) + if (__Pyx_IternextUnpackEndCheck(__pyx_t_9(__pyx_t_8), 2) < 0) __PYX_ERR(0, 461, __pyx_L1_error) __pyx_t_9 = NULL; __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; goto __pyx_L14_unpacking_done; @@ -3522,36 +3498,36 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0; __pyx_t_9 = NULL; if (__Pyx_IterFinish() == 0) __Pyx_RaiseNeedMoreValuesError(index); - __PYX_ERR(0, 463, __pyx_L1_error) + __PYX_ERR(0, 461, __pyx_L1_error) __pyx_L14_unpacking_done:; } - __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_7); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 463, __pyx_L1_error) + __pyx_t_10 = __Pyx_PyInt_As_npy_uint32(__pyx_t_7); if (unlikely((__pyx_t_10 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 461, __pyx_L1_error) __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0; - __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 463, __pyx_L1_error) + __pyx_t_2 = __Pyx_PyInt_As_npy_uint32(__pyx_t_6); if (unlikely((__pyx_t_2 == ((npy_uint32)-1)) && PyErr_Occurred())) __PYX_ERR(0, 461, __pyx_L1_error) __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0; __pyx_v_eff_sentences = __pyx_t_10; __pyx_v_eff_words = __pyx_t_2; - /* "pooling_inner.pyx":469 + /* "pooling_inner.pyx":467 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< * with nogil: * compute_ft_sentence_pooling( */ - __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 469, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_GetAttrStr(__pyx_v_model, __pyx_n_s_hierarchical); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 467, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 469, __pyx_L1_error) + __pyx_t_4 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_4 < 0)) __PYX_ERR(0, 467, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; __pyx_t_5 = ((!__pyx_t_4) != 0); if (__pyx_t_5) { - /* "pooling_inner.pyx":470 + /* "pooling_inner.pyx":468 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< * compute_ft_sentence_pooling( - * &ft, + * &config, */ { #ifdef WITH_THREAD @@ -3561,22 +3537,22 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":471 + /* "pooling_inner.pyx":469 * if not model.hierarchical: * with nogil: * compute_ft_sentence_pooling( # <<<<<<<<<<<<<< - * &ft, + * &config, * eff_sentences */ - __pyx_f_13pooling_inner_compute_ft_sentence_pooling((&__pyx_v_ft), __pyx_v_eff_sentences); + __pyx_f_13pooling_inner_compute_ft_sentence_pooling((&__pyx_v_config), __pyx_v_eff_sentences); } - /* "pooling_inner.pyx":470 + /* "pooling_inner.pyx":468 * * if not model.hierarchical: * with nogil: # <<<<<<<<<<<<<< * compute_ft_sentence_pooling( - * &ft, + * &config, */ /*finally:*/ { /*normal exit:*/{ @@ -3590,7 +3566,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } } - /* "pooling_inner.pyx":469 + /* "pooling_inner.pyx":467 * ) * * if not model.hierarchical: # <<<<<<<<<<<<<< @@ -3600,12 +3576,12 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec goto __pyx_L15; } - /* "pooling_inner.pyx":476 + /* "pooling_inner.pyx":474 * ) * else: * with nogil: # <<<<<<<<<<<<<< * compute_ft_sentence_hier_pooling( - * &ft, + * &config, */ /*else*/ { { @@ -3616,22 +3592,22 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec #endif /*try:*/ { - /* "pooling_inner.pyx":477 + /* "pooling_inner.pyx":475 * else: * with nogil: * compute_ft_sentence_hier_pooling( # <<<<<<<<<<<<<< - * &ft, + * &config, * eff_sentences, */ - __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling((&__pyx_v_ft), __pyx_v_eff_sentences, __pyx_v_window_size, __pyx_v_window_stride); + __pyx_f_13pooling_inner_compute_ft_sentence_hier_pooling((&__pyx_v_config), __pyx_v_eff_sentences, __pyx_v_window_size, __pyx_v_window_stride); } - /* "pooling_inner.pyx":476 + /* "pooling_inner.pyx":474 * ) * else: * with nogil: # <<<<<<<<<<<<<< * compute_ft_sentence_hier_pooling( - * &ft, + * &config, */ /*finally:*/ { /*normal exit:*/{ @@ -3649,7 +3625,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec } __pyx_L3:; - /* "pooling_inner.pyx":484 + /* "pooling_inner.pyx":482 * ) * * return eff_sentences, eff_words # <<<<<<<<<<<<<< @@ -3657,11 +3633,11 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec * def init(): */ __Pyx_XDECREF(__pyx_r); - __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 484, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_sentences); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 482, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); - __pyx_t_6 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 484, __pyx_L1_error) + __pyx_t_6 = __Pyx_PyInt_From_npy_uint32(__pyx_v_eff_words); if (unlikely(!__pyx_t_6)) __PYX_ERR(0, 482, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_6); - __pyx_t_7 = PyTuple_New(2); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 484, __pyx_L1_error) + __pyx_t_7 = PyTuple_New(2); if (unlikely(!__pyx_t_7)) __PYX_ERR(0, 482, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_7); __Pyx_GIVEREF(__pyx_t_1); PyTuple_SET_ITEM(__pyx_t_7, 0, __pyx_t_1); @@ -3673,7 +3649,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec __pyx_t_7 = 0; goto __pyx_L0; - /* "pooling_inner.pyx":401 + /* "pooling_inner.pyx":400 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< @@ -3695,7 +3671,7 @@ static PyObject *__pyx_pf_13pooling_inner_train_pooling_cy(CYTHON_UNUSED PyObjec return __pyx_r; } -/* "pooling_inner.pyx":486 +/* "pooling_inner.pyx":484 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -3723,7 +3699,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __Pyx_RefNannyDeclarations __Pyx_RefNannySetupContext("init", 0); - /* "pooling_inner.pyx":487 + /* "pooling_inner.pyx":485 * * def init(): * return 1 # <<<<<<<<<<<<<< @@ -3735,7 +3711,7 @@ static PyObject *__pyx_pf_13pooling_inner_2init(CYTHON_UNUSED PyObject *__pyx_se __pyx_r = __pyx_int_1; goto __pyx_L0; - /* "pooling_inner.pyx":486 + /* "pooling_inner.pyx":484 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< @@ -6225,10 +6201,10 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = { {&__pyx_n_s_ValueError, __pyx_k_ValueError, sizeof(__pyx_k_ValueError), 0, 0, 1, 1}, {&__pyx_n_s__8, __pyx_k__8, sizeof(__pyx_k__8), 0, 0, 1, 1}, {&__pyx_n_s_cline_in_traceback, __pyx_k_cline_in_traceback, sizeof(__pyx_k_cline_in_traceback), 0, 0, 1, 1}, + {&__pyx_n_s_config, __pyx_k_config, sizeof(__pyx_k_config), 0, 0, 1, 1}, {&__pyx_n_s_eff_sentences, __pyx_k_eff_sentences, sizeof(__pyx_k_eff_sentences), 0, 0, 1, 1}, {&__pyx_n_s_eff_words, __pyx_k_eff_words, sizeof(__pyx_k_eff_words), 0, 0, 1, 1}, {&__pyx_n_s_fblas, __pyx_k_fblas, sizeof(__pyx_k_fblas), 0, 0, 1, 1}, - {&__pyx_n_s_ft, __pyx_k_ft, sizeof(__pyx_k_ft), 0, 0, 1, 1}, {&__pyx_n_s_hierarchical, __pyx_k_hierarchical, sizeof(__pyx_k_hierarchical), 0, 0, 1, 1}, {&__pyx_n_s_import, __pyx_k_import, sizeof(__pyx_k_import), 0, 0, 1, 1}, {&__pyx_n_s_indexed_sentences, __pyx_k_indexed_sentences, sizeof(__pyx_k_indexed_sentences), 0, 0, 1, 1}, @@ -6253,14 +6229,13 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = { {&__pyx_n_s_train_pooling_cy, __pyx_k_train_pooling_cy, sizeof(__pyx_k_train_pooling_cy), 0, 0, 1, 1}, {&__pyx_kp_u_unknown_dtype_code_in_numpy_pxd, __pyx_k_unknown_dtype_code_in_numpy_pxd, sizeof(__pyx_k_unknown_dtype_code_in_numpy_pxd), 0, 1, 0, 0}, {&__pyx_n_s_vocab, __pyx_k_vocab, sizeof(__pyx_k_vocab), 0, 0, 1, 1}, - {&__pyx_n_s_w2v, __pyx_k_w2v, sizeof(__pyx_k_w2v), 0, 0, 1, 1}, {&__pyx_n_s_window_size, __pyx_k_window_size, sizeof(__pyx_k_window_size), 0, 0, 1, 1}, {&__pyx_n_s_window_stride, __pyx_k_window_stride, sizeof(__pyx_k_window_stride), 0, 0, 1, 1}, {&__pyx_n_s_wv, __pyx_k_wv, sizeof(__pyx_k_wv), 0, 0, 1, 1}, {0, 0, 0, 0, 0, 0, 0} }; static CYTHON_SMALL_CODE int __Pyx_InitCachedBuiltins(void) { - __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 96, __pyx_L1_error) + __pyx_builtin_range = __Pyx_GetBuiltinName(__pyx_n_s_range); if (!__pyx_builtin_range) __PYX_ERR(0, 95, __pyx_L1_error) __pyx_builtin_ValueError = __Pyx_GetBuiltinName(__pyx_n_s_ValueError); if (!__pyx_builtin_ValueError) __PYX_ERR(1, 272, __pyx_L1_error) __pyx_builtin_RuntimeError = __Pyx_GetBuiltinName(__pyx_n_s_RuntimeError); if (!__pyx_builtin_RuntimeError) __PYX_ERR(1, 856, __pyx_L1_error) __pyx_builtin_ImportError = __Pyx_GetBuiltinName(__pyx_n_s_ImportError); if (!__pyx_builtin_ImportError) __PYX_ERR(1, 1038, __pyx_L1_error) @@ -6350,26 +6325,26 @@ static CYTHON_SMALL_CODE int __Pyx_InitCachedConstants(void) { __Pyx_GOTREF(__pyx_tuple__7); __Pyx_GIVEREF(__pyx_tuple__7); - /* "pooling_inner.pyx":401 + /* "pooling_inner.pyx":400 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_tuple__9 = PyTuple_Pack(10, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_window_size, __pyx_n_s_window_stride, __pyx_n_s_w2v, __pyx_n_s_ft); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 401, __pyx_L1_error) + __pyx_tuple__9 = PyTuple_Pack(9, __pyx_n_s_model, __pyx_n_s_indexed_sentences, __pyx_n_s_target, __pyx_n_s_memory, __pyx_n_s_eff_sentences, __pyx_n_s_eff_words, __pyx_n_s_window_size, __pyx_n_s_window_stride, __pyx_n_s_config); if (unlikely(!__pyx_tuple__9)) __PYX_ERR(0, 400, __pyx_L1_error) __Pyx_GOTREF(__pyx_tuple__9); __Pyx_GIVEREF(__pyx_tuple__9); - __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 10, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 401, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 401, __pyx_L1_error) + __pyx_codeobj__10 = (PyObject*)__Pyx_PyCode_New(4, 0, 9, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_tuple__9, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_train_pooling_cy, 400, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__10)) __PYX_ERR(0, 400, __pyx_L1_error) - /* "pooling_inner.pyx":486 + /* "pooling_inner.pyx":484 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 486, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 486, __pyx_L1_error) + __pyx_codeobj__11 = (PyObject*)__Pyx_PyCode_New(0, 0, 0, 0, CO_OPTIMIZED|CO_NEWLOCALS, __pyx_empty_bytes, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_empty_tuple, __pyx_kp_s_pooling_inner_pyx, __pyx_n_s_init, 484, __pyx_empty_bytes); if (unlikely(!__pyx_codeobj__11)) __PYX_ERR(0, 484, __pyx_L1_error) __Pyx_RefNannyFinishContext(); return 0; __pyx_L1_error:; @@ -6491,10 +6466,10 @@ static int __Pyx_modinit_function_import_code(void) { __Pyx_RefNannySetupContext("__Pyx_modinit_function_import_code", 0); /*--- Function import code ---*/ __pyx_t_1 = PyImport_ImportModule("average_inner"); if (!__pyx_t_1) __PYX_ERR(0, 1, __pyx_L1_error) - if (__Pyx_ImportFunction(__pyx_t_1, "init_base_s2v_config", (void (**)(void))&__pyx_f_13average_inner_init_base_s2v_config, "PyObject *(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) - if (__Pyx_ImportFunction(__pyx_t_1, "init_ft_s2v_config", (void (**)(void))&__pyx_f_13average_inner_init_ft_s2v_config, "PyObject *(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) - if (__Pyx_ImportFunction(__pyx_t_1, "populate_base_s2v_config", (void (**)(void))&__pyx_f_13average_inner_populate_base_s2v_config, "PyObject *(struct __pyx_t_13average_inner_BaseSentenceVecsConfig *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) - if (__Pyx_ImportFunction(__pyx_t_1, "populate_ft_s2v_config", (void (**)(void))&__pyx_f_13average_inner_populate_ft_s2v_config, "PyObject *(struct __pyx_t_13average_inner_FTSentenceVecsConfig *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportFunction(__pyx_t_1, "init_base_s2v_config", (void (**)(void))&__pyx_f_13average_inner_init_base_s2v_config, "PyObject *(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportFunction(__pyx_t_1, "init_ft_s2v_config", (void (**)(void))&__pyx_f_13average_inner_init_ft_s2v_config, "PyObject *(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportFunction(__pyx_t_1, "populate_base_s2v_config", (void (**)(void))&__pyx_f_13average_inner_populate_base_s2v_config, "PyObject *(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) + if (__Pyx_ImportFunction(__pyx_t_1, "populate_ft_s2v_config", (void (**)(void))&__pyx_f_13average_inner_populate_ft_s2v_config, "PyObject *(struct __pyx_t_13average_inner_VecsConfig *, PyObject *, PyObject *)") < 0) __PYX_ERR(0, 1, __pyx_L1_error) Py_DECREF(__pyx_t_1); __pyx_t_1 = 0; __Pyx_RefNannyFinishContext(); return 0; @@ -6732,58 +6707,58 @@ if (!__Pyx_RefNanny) { if (PyDict_SetItem(__pyx_d, __pyx_n_s_fblas, __pyx_t_2) < 0) __PYX_ERR(0, 22, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":401 + /* "pooling_inner.pyx":400 * # There's nothing to do here for many-to-one mappings * * def train_pooling_cy( # <<<<<<<<<<<<<< * model, * indexed_sentences, */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 401, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_1train_pooling_cy, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 400, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 401, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_train_pooling_cy, __pyx_t_2) < 0) __PYX_ERR(0, 400, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":486 + /* "pooling_inner.pyx":484 * return eff_sentences, eff_words * * def init(): # <<<<<<<<<<<<<< * return 1 * */ - __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 486, __pyx_L1_error) + __pyx_t_2 = PyCFunction_NewEx(&__pyx_mdef_13pooling_inner_3init, NULL, __pyx_n_s_pooling_inner); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 484, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 486, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_init, __pyx_t_2) < 0) __PYX_ERR(0, 484, __pyx_L1_error) __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - /* "pooling_inner.pyx":489 + /* "pooling_inner.pyx":487 * return 1 * * MAX_WORDS_IN_BATCH = MAX_WORDS # <<<<<<<<<<<<<< * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 489, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_WORDS_IN_BATCH, __pyx_int_10000) < 0) __PYX_ERR(0, 487, __pyx_L1_error) - /* "pooling_inner.pyx":490 + /* "pooling_inner.pyx":488 * * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS # <<<<<<<<<<<<<< * FAST_VERSION = init() */ - if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 490, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_MAX_NGRAMS_IN_BATCH, __pyx_int_40) < 0) __PYX_ERR(0, 488, __pyx_L1_error) - /* "pooling_inner.pyx":491 + /* "pooling_inner.pyx":489 * MAX_WORDS_IN_BATCH = MAX_WORDS * MAX_NGRAMS_IN_BATCH = MAX_NGRAMS * FAST_VERSION = init() # <<<<<<<<<<<<<< */ - __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 491, __pyx_L1_error) + __Pyx_GetModuleGlobalName(__pyx_t_2, __pyx_n_s_init); if (unlikely(!__pyx_t_2)) __PYX_ERR(0, 489, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_2); - __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 491, __pyx_L1_error) + __pyx_t_1 = __Pyx_PyObject_CallNoArg(__pyx_t_2); if (unlikely(!__pyx_t_1)) __PYX_ERR(0, 489, __pyx_L1_error) __Pyx_GOTREF(__pyx_t_1); __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0; - if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 491, __pyx_L1_error) + if (PyDict_SetItem(__pyx_d, __pyx_n_s_FAST_VERSION, __pyx_t_1) < 0) __PYX_ERR(0, 489, __pyx_L1_error) __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0; /* "pooling_inner.pyx":1 diff --git a/fse/test/test_iterator.py b/fse/test/test_iterator.py new file mode 100644 index 0000000..f866c90 --- /dev/null +++ b/fse/test/test_iterator.py @@ -0,0 +1,43 @@ +#!/usr/bin/env python +# -*- coding: utf-8 -*- + +# Author: Oliver Borchers +# Copyright (C) 2020 Oliver Borchers + +from fse.models.base_iterator import base_iterator +from fse.test.model_shared_imports import * + +from fse.models.average import Average + +class TestSentenceIterator(unittest.TestCase): + def setUp(self): + self.sentences = IndexedLineDocument(CORPUS) + self.model = Average(W2V_DET, lang_freq="en") + + self.model.prep.prepare_vectors( + sv=self.model.sv, total_sentences=len(ENUM_SENTENCES), update=False + ) + self.model.window_size=2 + self.model.window_stride=2 + + self.model._pre_train_calls() + self.mem = self.model._get_thread_working_mem() + + def test_iterator_w2v_det(self): + def window_merger(*args, **kwargs): pass + def sentence_merger(*args, **kwargs): pass + + it = base_iterator( + self.model, + indexed_sentences = ENUM_SENTENCES, + target = self.model.sv.vectors, + memory = self.mem, + window_merger=window_merger, + sentence_merger=sentence_merger, + ) + +if __name__ == "__main__": + logging.basicConfig( + format="%(asctime)s : %(levelname)s : %(message)s", level=logging.DEBUG + ) + unittest.main() From b680e83f74e60352f55a321a1718ae241bdcc099 Mon Sep 17 00:00:00 2001 From: OliverB Date: Wed, 5 Aug 2020 11:12:39 +0200 Subject: [PATCH 57/63] Removed comments --- fse/models/average_inner.pxd | 23 ----------------------- 1 file changed, 23 deletions(-) diff --git a/fse/models/average_inner.pxd b/fse/models/average_inner.pxd index 326cf9e..5f390c4 100644 --- a/fse/models/average_inner.pxd +++ b/fse/models/average_inner.pxd @@ -69,29 +69,6 @@ cdef struct VecsConfig: uINT_t subwords_idx_len[MAX_WORDS] uINT_t *subwords_idx -# cdef struct FTSentenceVecsConfig: -# int size, workers, min_n, max_n, bucket - -# REAL_t oov_weight - -# # Vectors -# REAL_t *mem -# REAL_t *mem2 -# REAL_t *word_vectors # Note: these will be the vocab vectors, not wv.vectors -# REAL_t *ngram_vectors -# REAL_t *word_weights - -# REAL_t *sentence_vectors - -# # REAL_t *work memory for summation? -# uINT_t word_indices[MAX_WORDS] -# uINT_t sent_adresses[MAX_WORDS] -# uINT_t sentence_boundary[MAX_WORDS + 1] - -# # For storing the oov items -# uINT_t subwords_idx_len[MAX_WORDS] -# uINT_t *subwords_idx - cdef init_base_s2v_config( VecsConfig *c, model, From 68643a3f40bb4f3ba1803233667545ce0221151d Mon Sep 17 00:00:00 2001 From: OliverB Date: Wed, 5 Aug 2020 15:00:32 +0200 Subject: [PATCH 58/63] todo changed --- fse/models/base_s2v.py | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/fse/models/base_s2v.py b/fse/models/base_s2v.py index 6e49c4d..57fa72a 100644 --- a/fse/models/base_s2v.py +++ b/fse/models/base_s2v.py @@ -145,7 +145,7 @@ def __init__( Immediate [ ] pre_inference_calls [ ] base_iterator: - [ ] Rewrite VecsConfig to combine FT & W2V + [X] Rewrite VecsConfig to combine FT & W2V [ ] Hierarchical Average [ ] Stride Parameter: https://adeshpande3.github.io/A-Beginner%27s-Guide-To-Understanding-Convolutional-Neural-Networks-Part-2/ [ ] Average: From 9fb4883b1a58be7a9e20be3db63b46288f7e1e61 Mon Sep 17 00:00:00 2001 From: OliverB Date: Wed, 5 Aug 2020 15:00:47 +0200 Subject: [PATCH 59/63] First iterator draft --- fse/models/base_iterator.py | 128 +++++++++++++++++++++++------------- fse/test/test_iterator.py | 20 ++++-- 2 files changed, 97 insertions(+), 51 deletions(-) diff --git a/fse/models/base_iterator.py b/fse/models/base_iterator.py index 8a18e9e..7cc03ce 100644 --- a/fse/models/base_iterator.py +++ b/fse/models/base_iterator.py @@ -26,8 +26,11 @@ def base_iterator( indexed_sentences: List[tuple], target: ndarray, memory: tuple, - window_merger : callable, - sentence_merger : callable, + sentence_length : callable, + window_kernel : callable, + window_scaler : callable, + sentence_kernel : callable, + sentence_scaler : callable, ) -> [int, int]: # """Training on a sequence of sentences and update the target ndarray. @@ -41,7 +44,7 @@ def base_iterator( # Parameters # ---------- # model : :class:`~fse.models.base_s2v.BaseSentence2VecModel` - # The BaseSentence2VecModel model instance. + # The BaseSentence2VecModel model instance or a child of it. # indexed_sentences : iterable of tuple # The sentences used to train the model. # target : ndarray @@ -60,16 +63,16 @@ def base_iterator( size = model.wv.vector_size vocab = model.wv.vocab + mem = memory[0] + mem2 = memory[1] + # Do not need ngram vectors here due to numpy ft ngram implementation + w_vectors = model.wv.vectors w_weights = model.word_weights - s_vectors = target is_ft = model.is_ft - mem = memory[0] - mem2 = memory[1] - window_size = model.window_size window_stride = model.window_stride @@ -79,23 +82,21 @@ def base_iterator( # I suspect this is because the wv.vectors are based on the averages of # wv.vectors_vocab + wv.vectors_ngrams, which will all point into very # similar directions. - max_ngrams = model.batch_ngrams - w_vectors = model.wv.vectors_vocab - ngram_vectors = model.wv.vectors_ngrams min_n = model.wv.min_n max_n = model.wv.max_n bucket = model.wv.bucket + max_ngrams = model.batch_ngrams oov_weight = np_amax(w_weights) + w_vectors = model.wv.vectors_vocab + ngram_vectors = model.wv.vectors_ngrams + eff_sentences, eff_words = 0, 0 - for obj in indexed_sentences: - sent = obj[0] - sent_adr = obj[1] - + for sent, sent_index in indexed_sentences: if not len(sent): + # Skip if sentence is empty, leaves vector empty continue - eff_sentences += 1 # In cython we know the length (-OOV) beforehand @@ -106,28 +107,28 @@ def base_iterator( if not sent_len: continue - # Number of windows to be encountered + # Number of windows in a sentence. Includes broken windows at the edge win_count = int(ceil(sent_len / window_size)) for word_index, _ in enumerate(sent): if word_index % window_stride != 0: continue - win_len = 0 mem.fill(0.) # mem for summation mem2.fill(0.) + for word in sent[word_index : word_index + window_size]: - eff_words += window_func( + eff_words += window_kernel( model, word, mem, + mem2, ) # -> mem # W2V will return 0&1, FT returns 1 - win_len += 1 # Herein the window will be merged (i.e., rescaled) - window_merger( + window_scaler( model, win_len, mem, @@ -135,13 +136,22 @@ def base_iterator( ) # mem -> mem2 # Partially constructs the sentence onto sv.vectors - sentence_merger( + sentence_kernel( sent_len, - mem2, - sent_adr, + sent_index, target, + mem, + mem2, ) + sentence_scaler( + sent_len, + sent_index, + target, + mem, + mem2, + ) + return eff_sentences, eff_words def sentence_length( @@ -156,40 +166,68 @@ def sentence_length( # Inefficient, but hey, its just the python version anyways return sum([1 if token in model.wv.vocab else 0 for token in sent]) -def window_func( +def window_kernel( model, word : str, mem : ndarray, + mem2 : ndarray, ) -> int: - """ Computes the word vectors for a word + """ Window kernel implements aggregation function for window. + Does the vector conversion. + All results will be stored in mem. """ - if word in model.wv.vocab: - word_index = model.wv.vocab[word].index - mem += model.wv.vectors[word_index] * model.word_weights[word_index] - return 1 - else: - if model.is_ft: - mem += get_ft_word_vector(word, model) * np_max(model.word_weights) - return 1 - else: - return 0 # Word not taken into account + return 1 -def window_merger( +def window_scaler( model, - win_len : int, + window_length : int, mem : ndarray, mem2 : ndarray, - ): - """ Average window merger. - Should implement functionality to merge temporary results from - mem to mem2 inplace. Depends on model architecture + ) -> None: + """ Window scaler implements scaling function for window result. + All results will be stored in mem2. """ pass -def sentence_merger( - window_length : int, +def sentence_kernel( + sent_length : int, + sent_index : int, + target : ndarray, mem : ndarray, + mem2 : ndarray, + ) -> int: + """ Sentence kernel implements aggregation function for all windows. + All results will be stored in target. + """ + pass + +def sentence_scaler( + sent_length : int, sent_adr : int, target : ndarray, - ): - pass \ No newline at end of file + mem : ndarray, + mem2 : ndarray, + ) -> None: + """ Sentence scaler implements scaling function for accumulated window result. + All results will be stored in target. + """ + pass + +# def window_func( +# model, +# word : str, +# mem : ndarray, +# ) -> int: +# """ Computes the word vectors for a word +# """ +# if word in model.wv.vocab: +# word_index = model.wv.vocab[word].index +# mem += model.wv.vectors[word_index] * model.word_weights[word_index] +# return 1 +# else: +# if model.is_ft: +# mem += get_ft_word_vector(word, model) * np_max(model.word_weights) +# return 1 +# else: +# return 0 # Word not taken into account + diff --git a/fse/test/test_iterator.py b/fse/test/test_iterator.py index f866c90..40b0a46 100644 --- a/fse/test/test_iterator.py +++ b/fse/test/test_iterator.py @@ -4,7 +4,15 @@ # Author: Oliver Borchers # Copyright (C) 2020 Oliver Borchers -from fse.models.base_iterator import base_iterator +from fse.models.base_iterator import ( + base_iterator, + sentence_length, + window_kernel, + window_scaler, + sentence_kernel, + sentence_scaler, +) + from fse.test.model_shared_imports import * from fse.models.average import Average @@ -24,16 +32,16 @@ def setUp(self): self.mem = self.model._get_thread_working_mem() def test_iterator_w2v_det(self): - def window_merger(*args, **kwargs): pass - def sentence_merger(*args, **kwargs): pass - it = base_iterator( self.model, indexed_sentences = ENUM_SENTENCES, target = self.model.sv.vectors, memory = self.mem, - window_merger=window_merger, - sentence_merger=sentence_merger, + sentence_length=sentence_length, + window_kernel=window_kernel, + window_scaler=window_scaler, + sentence_kernel=sentence_kernel, + sentence_scaler=sentence_scaler, ) if __name__ == "__main__": From f858274341de4ca80cf199be6f873458dbf6f917 Mon Sep 17 00:00:00 2001 From: OliverB Date: Wed, 5 Aug 2020 15:13:58 +0200 Subject: [PATCH 60/63] Added option to skip Cython tests --- fse/test/model_shared_imports.py | 2 + fse/test/test_average.py | 8 +- fse/test/test_base_s2v.py | 1 + fse/test/test_pooling.py | 953 +++++++++++++++---------------- 4 files changed, 484 insertions(+), 480 deletions(-) diff --git a/fse/test/model_shared_imports.py b/fse/test/model_shared_imports.py index e76dc4f..a3373b0 100644 --- a/fse/test/model_shared_imports.py +++ b/fse/test/model_shared_imports.py @@ -18,6 +18,8 @@ SENTENCES = [l.split() for i, l in enumerate(f)] ENUM_SENTENCES = [(l, i) for i, l in enumerate(SENTENCES)] +IGNORE_CY = True + # Models DIM = 5 diff --git a/fse/test/test_average.py b/fse/test/test_average.py index 862daf0..d5bd454 100644 --- a/fse/test/test_average.py +++ b/fse/test/test_average.py @@ -28,6 +28,7 @@ def setUp(self): ) self.model._pre_train_calls() + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") def test_cython(self): from fse.models.average_inner import ( FAST_VERSION, @@ -49,7 +50,8 @@ def test_average_train_np_w2v_det(self): self.assertTrue((183 == self.model.sv[0]).all()) self.assertTrue((164.5 == self.model.sv[1]).all()) self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) - + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") def test_average_train_cy_w2v_det(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() @@ -81,6 +83,7 @@ def test_average_train_np_ft_det(self): # oov: "12345" -> (14 hashes * 2) / 14 = 2 # (2 + 1) / 2 = 1.5 + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") def test_average_train_cy_ft_det(self): m = Average(FT_DET) m.prep.prepare_vectors( @@ -97,6 +100,7 @@ def test_average_train_cy_ft_det(self): self.assertTrue(np.allclose(368707.4, m.sv[2])) self.assertTrue(np.allclose(961940.0, m.sv[3])) + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") def test_cy_equal_np_w2v_det(self): m1 = Average(W2V_DET) m1.prep.prepare_vectors( @@ -120,6 +124,7 @@ def test_cy_equal_np_w2v_det(self): self.assertEqual(o1, o2) self.assertTrue((m1.sv.vectors == m2.sv.vectors).all()) + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") def test_cy_equal_np_w2v_rng(self): m1 = Average(W2V_RNG) m1.prep.prepare_vectors( @@ -142,6 +147,7 @@ def test_cy_equal_np_w2v_rng(self): self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") def test_cy_equal_np_ft_rng(self): m1 = Average(FT_RNG) m1.prep.prepare_vectors( diff --git a/fse/test/test_base_s2v.py b/fse/test/test_base_s2v.py index 7df4043..5334601 100644 --- a/fse/test/test_base_s2v.py +++ b/fse/test/test_base_s2v.py @@ -454,6 +454,7 @@ def pass_method(**kwargs): output = se.infer([(s, i) for i, s in enumerate(SENTENCES)]) self.assertTrue((100 == output).all()) + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") def test_infer_method_cy_overflow(self): se = BaseSentence2VecModel(W2V) diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index 1ec9cb0..869f794 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -25,493 +25,488 @@ def setUp(self): ) self.model._pre_train_calls() - # def test_cython(self): - # from fse.models.pooling_inner import ( - # FAST_VERSION, - # MAX_WORDS_IN_BATCH, - # MAX_NGRAMS_IN_BATCH, - # train_pooling_cy, - # ) - - # self.assertTrue(FAST_VERSION) - # self.assertTrue(callable(train_pooling_cy)) - # self.assertEqual(10000, MAX_WORDS_IN_BATCH) - # self.assertEqual(40, MAX_NGRAMS_IN_BATCH) - - # def test_check_parameter_sanity(self): - # se = MaxPooling(W2V) - # se.word_weights = np.full(20, 2.0, dtype=np.float32) - # with self.assertRaises(ValueError): - # se._check_parameter_sanity() - - # se = MaxPooling(W2V, window_size=0) - # with self.assertRaises(ValueError): - # se._check_parameter_sanity() - - # se = MaxPooling(W2V, window_size=3, window_stride=0) - # with self.assertRaises(ValueError): - # se._check_parameter_sanity() - - # se = MaxPooling(W2V, window_size=3, window_stride=4) - # with self.assertRaises(ValueError): - # se._check_parameter_sanity() - - # def test_train(self): - # self.assertEqual( - # (100, 1450), self.model.train([(s, i) for i, s in enumerate(SENTENCES)]) - # ) - - # def test_do_train_job(self): - # self.model.prep.prepare_vectors( - # sv=self.model.sv, total_sentences=len(SENTENCES), update=True - # ) - # mem = self.model._get_thread_working_mem() - # self.assertEqual( - # (100, 1450), - # self.model._do_train_job( - # [(s, i) for i, s in enumerate(SENTENCES)], - # target=self.model.sv.vectors, - # memory=mem, - # ), - # ) - # self.assertEqual((105, DIM), self.model.sv.vectors.shape) - - # ### Basic Pooling Tests start here - - # def test_pool_train_np_w2v(self): - # self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) - # mem = self.model._get_thread_working_mem() - - # output = train_pooling_np( - # self.model, self.sentences, self.model.sv.vectors, mem - # ) - - # self.assertEqual((5, 14), output) - # self.assertTrue((241 == self.model.sv[0]).all()) - # self.assertTrue((306 == self.model.sv[1]).all()) - # self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) - - # def test_pool_train_cy_w2v(self): - # self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) - # mem = self.model._get_thread_working_mem() - - # from fse.models.pooling_inner import train_pooling_cy - - # output = train_pooling_cy( - # self.model, self.sentences, self.model.sv.vectors, mem - # ) - - # self.assertEqual((5, 14), output) - # self.assertTrue((241 == self.model.sv[0]).all()) - # self.assertTrue((306 == self.model.sv[1]).all()) - # self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) - - -###### Worked until here - -# def test_pool_train_np_ft(self): -# m = MaxPooling(FT) -# m.prep.prepare_vectors( -# sv=m.sv, total_sentences=len(self.sentences), update=False -# ) -# m._pre_train_calls() -# mem = m._get_thread_working_mem() - -# output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) - -# self.assertEqual((5, 19), output) -# self.assertTrue((241 == m.sv[0]).all()) -# self.assertTrue( -# np.allclose(737413.9, m.sv[2]) -# ) -# self.assertTrue( -# np.allclose(1080970.2, m.sv[3]) -# ) - -# def test_pool_train_cy_ft(self): -# m = MaxPooling(FT) -# m.prep.prepare_vectors( -# sv=m.sv, total_sentences=len(self.sentences), update=False -# ) -# m._pre_train_calls() -# mem = m._get_thread_working_mem() - -# from fse.models.pooling_inner import train_pooling_cy - -# output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) - -# self.assertEqual((5, 19), output) -# self.assertTrue((241 == m.sv[0]).all()) -# self.assertTrue( -# np.allclose(737413.9, m.sv[2]) -# ) -# self.assertTrue( -# np.allclose(1080970.2, m.sv[3]) -# ) - -# def test_pool_cy_equal_np_w2v(self): -# m1 = MaxPooling(W2V) -# m1.prep.prepare_vectors( -# sv=m1.sv, total_sentences=len(self.sentences), update=False -# ) -# m1._pre_train_calls() -# mem1 = m1._get_thread_working_mem() -# o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) - -# m2 = MaxPooling(W2V) -# m2.prep.prepare_vectors( -# sv=m2.sv, total_sentences=len(self.sentences), update=False -# ) -# m2._pre_train_calls() -# mem2 = m2._get_thread_working_mem() - -# from fse.models.pooling_inner import train_pooling_cy - -# o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) - -# self.assertEqual(o1, o2) -# self.assertTrue((m1.sv.vectors == m2.sv.vectors).all()) - -# def test_pool_cy_equal_np_w2v_random(self): -# w2v = Word2Vec(min_count=1, size=DIM) -# # Random initialization -# w2v.build_vocab(SENTENCES) - -# m1 = MaxPooling(w2v) -# m1.prep.prepare_vectors( -# sv=m1.sv, total_sentences=len(self.sentences), update=False -# ) -# m1._pre_train_calls() -# mem1 = m1._get_thread_working_mem() -# o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) - -# m2 = MaxPooling(w2v) -# m2.prep.prepare_vectors( -# sv=m2.sv, total_sentences=len(self.sentences), update=False -# ) -# m2._pre_train_calls() -# mem2 = m2._get_thread_working_mem() - -# from fse.models.pooling_inner import train_pooling_cy - -# o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) - -# self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - -# def test_pool_cy_equal_np_ft_random(self): -# ft = FastText(size=20, min_count=1) -# ft.build_vocab(SENTENCES) - -# m1 = MaxPooling(ft) -# m1.prep.prepare_vectors( -# sv=m1.sv, total_sentences=len(self.sentences), update=False -# ) -# m1._pre_train_calls() - -# from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH - -# m1.batch_ngrams = MAX_NGRAMS_IN_BATCH -# mem1 = m1._get_thread_working_mem() -# o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - -# m2 = MaxPooling(ft) -# m2.prep.prepare_vectors( -# sv=m2.sv, total_sentences=len(self.sentences), update=False -# ) -# m2._pre_train_calls() -# mem2 = m2._get_thread_working_mem() - -# from fse.models.pooling_inner import train_pooling_cy - -# o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) - -# self.assertEqual(o1, o2) -# self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - -# def test_pool_np_w2v_non_negative(self): -# mpool = MaxPooling(W2V_R) -# mpool.train(self.sentences) -# self.assertTrue((mpool.sv.vectors >= 0).all()) - -# def test_pool_np_ft_non_negative(self): -# mpool = MaxPooling(FT_R) -# mpool.train(self.sentences) -# self.assertTrue((mpool.sv.vectors >= 0).all()) - -### Hierarchical Tests start here - -# def test_hier_pool_train_np_w2v(self): -# self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) -# mem = self.model._get_thread_working_mem() - -# self.model.hierarchical = True - -# output = train_pooling_np( -# self.model, self.sentences, self.model.sv.vectors, mem -# ) -# self.model.hierarchical = False - -# self.assertEqual((5, 14), output) -# self.assertTrue((183 == self.model.sv[0]).all()) -# self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) - -# def test_hier_pool_train_cy_w2v(self): -# self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) -# mem = self.model._get_thread_working_mem() - -# self.model.hierarchical = True - -# from fse.models.pooling_inner import train_pooling_cy - -# output = train_pooling_cy( -# self.model, self.sentences, self.model.sv.vectors, mem -# ) -# self.model.hierarchical = False - -# self.assertEqual((5, 14), output) -# self.assertTrue((183 == self.model.sv[0]).all()) -# self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) - -# def test_hier_pool_train_np_ft(self): -# m = MaxPooling(FT, hierarchical=True) -# m.prep.prepare_vectors( -# sv=m.sv, total_sentences=len(self.sentences), update=False -# ) -# m._pre_train_calls() -# mem = m._get_thread_working_mem() - -# output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) - -# self.assertEqual((5, 19), output) -# self.assertTrue((183 == m.sv[0]).all()) -# self.assertTrue(np.allclose(737413.9, m.sv[2])) -# self.assertTrue(np.allclose(1080970.2, m.sv[3])) -# """ -# Note to future self: -# Due to the size of the ngram vectors, -# an ngram at the last position of the senence -# will always be the highest value. -# TODO: This unittest is thus a bit flawed. Maybe fix? -# """ + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_cython(self): + from fse.models.pooling_inner import ( + FAST_VERSION, + MAX_WORDS_IN_BATCH, + MAX_NGRAMS_IN_BATCH, + train_pooling_cy, + ) + + self.assertTrue(FAST_VERSION) + self.assertTrue(callable(train_pooling_cy)) + self.assertEqual(10000, MAX_WORDS_IN_BATCH) + self.assertEqual(40, MAX_NGRAMS_IN_BATCH) + + def test_check_parameter_sanity(self): + se = MaxPooling(W2V_DET) + se.word_weights = np.full(20, 2.0, dtype=np.float32) + with self.assertRaises(ValueError): + se._check_parameter_sanity() + + se = MaxPooling(W2V_DET, window_size=0) + with self.assertRaises(ValueError): + se._check_parameter_sanity() + + se = MaxPooling(W2V_DET, window_size=3, window_stride=0) + with self.assertRaises(ValueError): + se._check_parameter_sanity() + + se = MaxPooling(W2V_DET, window_size=3, window_stride=4) + with self.assertRaises(ValueError): + se._check_parameter_sanity() + + def test_train(self): + self.assertEqual( + (100, 1450), self.model.train([(s, i) for i, s in enumerate(SENTENCES)]) + ) + + def test_do_train_job(self): + self.model.prep.prepare_vectors( + sv=self.model.sv, total_sentences=len(SENTENCES), update=True + ) + mem = self.model._get_thread_working_mem() + self.assertEqual( + (100, 1450), + self.model._do_train_job( + [(s, i) for i, s in enumerate(SENTENCES)], + target=self.model.sv.vectors, + memory=mem, + ), + ) + self.assertEqual((105, DIM), self.model.sv.vectors.shape) + + ### Basic Pooling Tests start here + + def test_pool_train_np_w2v(self): + self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + mem = self.model._get_thread_working_mem() + + output = train_pooling_np( + self.model, self.sentences, self.model.sv.vectors, mem + ) + + self.assertEqual((5, 14), output) + self.assertTrue((241 == self.model.sv[0]).all()) + self.assertTrue((306 == self.model.sv[1]).all()) + self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_pool_train_cy_w2v(self): + self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + mem = self.model._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + output = train_pooling_cy( + self.model, self.sentences, self.model.sv.vectors, mem + ) + + self.assertEqual((5, 14), output) + self.assertTrue((241 == self.model.sv[0]).all()) + self.assertTrue((306 == self.model.sv[1]).all()) + self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) + + + ##### Worked until here + + def test_pool_train_np_ft(self): + m = MaxPooling(FT_DET) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + mem = m._get_thread_working_mem() + + output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + self.assertTrue((1. == m.sv[0]).all()) + self.assertTrue( + np.allclose(737413.9, m.sv[2]) + ) + self.assertTrue( + np.allclose(1080970.2, m.sv[3]) + ) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_pool_train_cy_ft(self): + m = MaxPooling(FT_DET) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + mem = m._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy -# def test_hier_pool_train_cy_ft(self): -# m = MaxPooling(FT, hierarchical=True) -# m.prep.prepare_vectors( -# sv=m.sv, total_sentences=len(self.sentences), update=False -# ) -# m._pre_train_calls() -# mem = m._get_thread_working_mem() - -# from fse.models.pooling_inner import train_pooling_cy - -# output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) + output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) -# self.assertEqual((5, 19), output) -# self.assertTrue((183 == m.sv[0]).all()) -# self.assertTrue(np.allclose(737413.9, m.sv[2])) -# self.assertTrue(np.allclose(1080970.2, m.sv[3])) - -# def test_hier_pool_cy_equal_np_w2v_random(self): -# w2v = Word2Vec(min_count=1, size=DIM) -# # Random initialization -# w2v.build_vocab(SENTENCES) - -# m1 = MaxPooling(w2v, hierarchical=True) -# m1.prep.prepare_vectors( -# sv=m1.sv, total_sentences=len(self.sentences), update=False -# ) -# m1._pre_train_calls() -# mem1 = m1._get_thread_working_mem() -# o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) - -# m2 = MaxPooling(w2v, hierarchical=True) -# m2.prep.prepare_vectors( -# sv=m2.sv, total_sentences=len(self.sentences), update=False -# ) -# m2._pre_train_calls() -# mem2 = m2._get_thread_working_mem() - -# from fse.models.pooling_inner import train_pooling_cy - -# o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) - -# self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - -# def test_hier_pool_cy_equal_np_ft_random(self): -# ft = FastText(size=20, min_count=1) -# ft.build_vocab(SENTENCES) - -# m1 = MaxPooling(ft, hierarchical=True) -# m1.prep.prepare_vectors( -# sv=m1.sv, total_sentences=len(self.sentences), update=False -# ) -# m1._pre_train_calls() - -# from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH - -# m1.batch_ngrams = MAX_NGRAMS_IN_BATCH -# mem1 = m1._get_thread_working_mem() -# o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - -# m2 = MaxPooling(ft, hierarchical=True) -# m2.prep.prepare_vectors( -# sv=m2.sv, total_sentences=len(self.sentences), update=False -# ) -# m2._pre_train_calls() -# mem2 = m2._get_thread_working_mem() - -# from fse.models.pooling_inner import train_pooling_cy - -# o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) - -# self.assertEqual(o1, o2) -# self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - -# def test_hier_pool_cy_w2v_non_negative(self): -# mpool = MaxPooling(W2V_R, hierarchical=True) -# mpool.train(self.sentences) -# self.assertTrue((mpool.sv.vectors >= 0).all()) - -# def test_hier_pool_cy_ft_non_negative(self): -# mpool = MaxPooling(FT_R, hierarchical=True) -# mpool.train(self.sentences) -# self.assertTrue((mpool.sv.vectors >= 0).all()) - -# ### Hierarchical Test + Stride start here - -# def test_hier_pool_stride_train_np_w2v(self): -# self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) -# mem = self.model._get_thread_working_mem() - -# self.model.hierarchical = True -# self.model.window_stride = 5 - -# output = train_pooling_np( -# self.model, self.sentences, self.model.sv.vectors, mem -# ) -# self.model.hierarchical = False -# self.model.window_stride = 1 - -# self.assertEqual((5, 14), output) -# self.assertTrue((183 == self.model.sv[0]).all()) -# self.assertTrue((231 == self.model.sv[4]).all()) - -# def test_hier_pool_stride_train_cy_w2v(self): -# self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) -# mem = self.model._get_thread_working_mem() + self.assertEqual((5, 19), output) + self.assertTrue((1. == m.sv[0]).all()) + self.assertTrue( + np.allclose(737413.9, m.sv[2]) + ) + self.assertTrue( + np.allclose(1080970.2, m.sv[3]) + ) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_pool_cy_equal_np_w2v(self): + m1 = MaxPooling(W2V_DET) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + + m2 = MaxPooling(W2V_DET) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + + self.assertEqual(o1, o2) + self.assertTrue((m1.sv.vectors == m2.sv.vectors).all()) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_pool_cy_equal_np_w2v_random(self): + m1 = MaxPooling(W2V_RNG) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + + m2 = MaxPooling(W2V_RNG) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + + self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_pool_cy_equal_np_ft_random(self): + m1 = MaxPooling(FT_RNG) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + + from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH + + m1.batch_ngrams = MAX_NGRAMS_IN_BATCH + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) + + m2 = MaxPooling(FT_RNG) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) + + self.assertEqual(o1, o2) + self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + def test_pool_np_w2v_non_negative(self): + mpool = MaxPooling(W2V_RNG) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) + + def test_pool_np_ft_non_negative(self): + mpool = MaxPooling(FT_RNG) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) + + ## Hierarchical Tests start here + + def test_hier_pool_train_np_w2v(self): + self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + mem = self.model._get_thread_working_mem() + + self.model.hierarchical = True + + output = train_pooling_np( + self.model, self.sentences, self.model.sv.vectors, mem + ) + self.model.hierarchical = False -# self.model.hierarchical = True -# self.model.window_stride = 5 + self.assertEqual((5, 14), output) + self.assertTrue((183 == self.model.sv[0]).all()) + self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_hier_pool_train_cy_w2v(self): + self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + mem = self.model._get_thread_working_mem() + + self.model.hierarchical = True + + from fse.models.pooling_inner import train_pooling_cy + + output = train_pooling_cy( + self.model, self.sentences, self.model.sv.vectors, mem + ) + self.model.hierarchical = False + + self.assertEqual((5, 14), output) + self.assertTrue((183 == self.model.sv[0]).all()) + self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) + + def test_hier_pool_train_np_ft(self): + m = MaxPooling(FT_DET, hierarchical=True) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + mem = m._get_thread_working_mem() + + output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + self.assertTrue((1 == m.sv[0]).all()) + self.assertTrue(np.allclose(737413.9, m.sv[2])) + self.assertTrue(np.allclose(1080970.2, m.sv[3])) + """ + Note to future self: + Due to the size of the ngram vectors, + an ngram at the last position of the senence + will always be the highest value. + TODO: This unittest is thus a bit flawed. Maybe fix? + """ + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_hier_pool_train_cy_ft(self): + m = MaxPooling(FT_DET, hierarchical=True) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + mem = m._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + self.assertTrue((1 == m.sv[0]).all()) + self.assertTrue(np.allclose(737413.9, m.sv[2])) + self.assertTrue(np.allclose(1080970.2, m.sv[3])) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_hier_pool_cy_equal_np_w2v_random(self): + m1 = MaxPooling(W2V_RNG, hierarchical=True) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + + m2 = MaxPooling(W2V_RNG, hierarchical=True) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + + self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_hier_pool_cy_equal_np_ft_random(self): + m1 = MaxPooling(FT_RNG, hierarchical=True) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + + from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH + + m1.batch_ngrams = MAX_NGRAMS_IN_BATCH + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) + + m2 = MaxPooling(FT_RNG, hierarchical=True) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) + + self.assertEqual(o1, o2) + self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_hier_pool_cy_w2v_non_negative(self): + mpool = MaxPooling(W2V_RNG, hierarchical=True) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_hier_pool_cy_ft_non_negative(self): + mpool = MaxPooling(FT_RNG, hierarchical=True) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) + + ### Hierarchical Test + Stride start here + + def test_hier_pool_stride_train_np_w2v(self): + self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + mem = self.model._get_thread_working_mem() + + self.model.hierarchical = True + self.model.window_stride = 5 + + output = train_pooling_np( + self.model, self.sentences, self.model.sv.vectors, mem + ) + self.model.hierarchical = False + self.model.window_stride = 1 + + self.assertEqual((5, 14), output) + self.assertTrue((183 == self.model.sv[0]).all()) + self.assertTrue((231 == self.model.sv[4]).all()) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_hier_pool_stride_train_cy_w2v(self): + self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) + mem = self.model._get_thread_working_mem() + + self.model.hierarchical = True + self.model.window_stride = 5 + + from fse.models.pooling_inner import train_pooling_cy + + output = train_pooling_cy( + self.model, self.sentences, self.model.sv.vectors, mem + ) + self.model.hierarchical = False + self.model.window_stride = 1 + + self.assertEqual((5, 14), output) + self.assertTrue((183 == self.model.sv[0]).all()) + self.assertTrue((231 == self.model.sv[4]).all()) + + def test_hier_pool_stride_train_np_ft(self): + m = MaxPooling(FT_DET, hierarchical=True, window_stride=3) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + mem = m._get_thread_working_mem() + + output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + self.assertTrue((1 == m.sv[0]).all()) + self.assertTrue(np.allclose(368707.44, m.sv[2])) + self.assertTrue(np.allclose(961940.2, m.sv[3])) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_hier_pool_stride_train_cy_ft(self): + m = MaxPooling(FT_DET, hierarchical=True, window_stride=3) + m.prep.prepare_vectors( + sv=m.sv, total_sentences=len(self.sentences), update=False + ) + m._pre_train_calls() + mem = m._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) + + self.assertEqual((5, 19), output) + self.assertTrue((1 == m.sv[0]).all()) + self.assertTrue(np.allclose(368707.44, m.sv[2])) + self.assertTrue(np.allclose(961940.2, m.sv[3])) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_hier_pool_stride_cy_equal_np_w2v_random(self): + m1 = MaxPooling(W2V_RNG, hierarchical=True, window_stride=4) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + + m2 = MaxPooling(W2V_RNG, hierarchical=True, window_stride=4) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() + + from fse.models.pooling_inner import train_pooling_cy + + o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) + + self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") + def test_hier_pool_stride_cy_equal_np_ft_random(self): + m1 = MaxPooling(FT_RNG, hierarchical=True, window_stride=5) + m1.prep.prepare_vectors( + sv=m1.sv, total_sentences=len(self.sentences), update=False + ) + m1._pre_train_calls() + + from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH + + m1.batch_ngrams = MAX_NGRAMS_IN_BATCH + mem1 = m1._get_thread_working_mem() + o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) + + m2 = MaxPooling(FT_RNG, hierarchical=True, window_stride=5) + m2.prep.prepare_vectors( + sv=m2.sv, total_sentences=len(self.sentences), update=False + ) + m2._pre_train_calls() + mem2 = m2._get_thread_working_mem() -# from fse.models.pooling_inner import train_pooling_cy - -# output = train_pooling_cy( -# self.model, self.sentences, self.model.sv.vectors, mem -# ) -# self.model.hierarchical = False -# self.model.window_stride = 1 - -# self.assertEqual((5, 14), output) -# self.assertTrue((183 == self.model.sv[0]).all()) -# self.assertTrue((231 == self.model.sv[4]).all()) - -# def test_hier_pool_stride_train_np_ft(self): -# m = MaxPooling(FT, hierarchical=True, window_stride=3) -# m.prep.prepare_vectors( -# sv=m.sv, total_sentences=len(self.sentences), update=False -# ) -# m._pre_train_calls() -# mem = m._get_thread_working_mem() + from fse.models.pooling_inner import train_pooling_cy -# output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) - -# self.assertEqual((5, 19), output) -# self.assertTrue((183 == m.sv[0]).all()) -# self.assertTrue(np.allclose(368871.94, m.sv[2])) -# self.assertTrue(np.allclose(961940.2, m.sv[3])) + o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) -# def test_hier_pool_stride_train_cy_ft(self): -# m = MaxPooling(FT, hierarchical=True, window_stride=3) -# m.prep.prepare_vectors( -# sv=m.sv, total_sentences=len(self.sentences), update=False -# ) -# m._pre_train_calls() -# mem = m._get_thread_working_mem() - -# from fse.models.pooling_inner import train_pooling_cy - -# output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) - -# self.assertEqual((5, 19), output) -# self.assertTrue((183 == m.sv[0]).all()) -# self.assertTrue(np.allclose(368871.94, m.sv[2])) -# self.assertTrue(np.allclose(961940.2, m.sv[3])) + self.assertEqual(o1, o2) + self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) -# def test_hier_pool_stride_cy_equal_np_w2v_random(self): -# w2v = Word2Vec(min_count=1, size=DIM) -# # Random initialization -# w2v.build_vocab(SENTENCES) - -# m1 = MaxPooling(w2v, hierarchical=True, window_stride=4) -# m1.prep.prepare_vectors( -# sv=m1.sv, total_sentences=len(self.sentences), update=False -# ) -# m1._pre_train_calls() -# mem1 = m1._get_thread_working_mem() -# o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) + def test_hier_pool_stride_np_w2v_non_negative(self): + mpool = MaxPooling(W2V_RNG, hierarchical=True, window_stride=4) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) -# m2 = MaxPooling(w2v, hierarchical=True, window_stride=4) -# m2.prep.prepare_vectors( -# sv=m2.sv, total_sentences=len(self.sentences), update=False -# ) -# m2._pre_train_calls() -# mem2 = m2._get_thread_working_mem() - -# from fse.models.pooling_inner import train_pooling_cy - -# o2 = train_pooling_cy(m2, self.sentences, m2.sv.vectors, mem2) - -# self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - -# def test_hier_pool_stride_cy_equal_np_ft_random(self): -# ft = FastText(size=20, min_count=1) -# ft.build_vocab(SENTENCES) - -# m1 = MaxPooling(ft, hierarchical=True, window_stride=5) -# m1.prep.prepare_vectors( -# sv=m1.sv, total_sentences=len(self.sentences), update=False -# ) -# m1._pre_train_calls() - -# from fse.models.pooling_inner import MAX_NGRAMS_IN_BATCH - -# m1.batch_ngrams = MAX_NGRAMS_IN_BATCH -# mem1 = m1._get_thread_working_mem() -# o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - -# m2 = MaxPooling(ft, hierarchical=True, window_stride=5) -# m2.prep.prepare_vectors( -# sv=m2.sv, total_sentences=len(self.sentences), update=False -# ) -# m2._pre_train_calls() -# mem2 = m2._get_thread_working_mem() - -# from fse.models.pooling_inner import train_pooling_cy - -# o2 = train_pooling_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) - -# self.assertEqual(o1, o2) -# self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - -# def test_hier_pool_stride_np_w2v_non_negative(self): -# mpool = MaxPooling(W2V_R, hierarchical=True, window_stride=4) -# mpool.train(self.sentences) -# self.assertTrue((mpool.sv.vectors >= 0).all()) - -# def test_hier_pool_stride_np_ft_non_negative(self): -# mpool = MaxPooling(FT_R, hierarchical=True, window_stride=4) -# mpool.train(self.sentences) -# self.assertTrue((mpool.sv.vectors >= 0).all()) + def test_hier_pool_stride_np_ft_non_negative(self): + mpool = MaxPooling(FT_RNG, hierarchical=True, window_stride=4) + mpool.train(self.sentences) + self.assertTrue((mpool.sv.vectors >= 0).all()) if __name__ == "__main__": From 6d20af0fdc4d8e7f5ff0b41523409a470b156800 Mon Sep 17 00:00:00 2001 From: OliverB Date: Thu, 6 Aug 2020 10:54:31 +0200 Subject: [PATCH 61/63] Python base_iterator + numpy pooling working --- fse/models/base_iterator.py | 29 +-- fse/models/pooling.py | 451 +++++++++++++++++++++--------------- fse/test/test_pooling.py | 163 ++++++++----- 3 files changed, 378 insertions(+), 265 deletions(-) diff --git a/fse/models/base_iterator.py b/fse/models/base_iterator.py index 7cc03ce..143ec8d 100644 --- a/fse/models/base_iterator.py +++ b/fse/models/base_iterator.py @@ -108,7 +108,7 @@ def base_iterator( continue # Number of windows in a sentence. Includes broken windows at the edge - win_count = int(ceil(sent_len / window_size)) + win_count = int(ceil(sent_len / window_stride)) for word_index, _ in enumerate(sent): if word_index % window_stride != 0: @@ -144,8 +144,10 @@ def base_iterator( mem2, ) + # Rescales the sentence if necessary + # Note: If size & stride = 1 -> win_count = sent_len sentence_scaler( - sent_len, + win_count, sent_index, target, mem, @@ -202,7 +204,7 @@ def sentence_kernel( pass def sentence_scaler( - sent_length : int, + win_count : int, sent_adr : int, target : ndarray, mem : ndarray, @@ -211,23 +213,4 @@ def sentence_scaler( """ Sentence scaler implements scaling function for accumulated window result. All results will be stored in target. """ - pass - -# def window_func( -# model, -# word : str, -# mem : ndarray, -# ) -> int: -# """ Computes the word vectors for a word -# """ -# if word in model.wv.vocab: -# word_index = model.wv.vocab[word].index -# mem += model.wv.vectors[word_index] * model.word_weights[word_index] -# return 1 -# else: -# if model.is_ft: -# mem += get_ft_word_vector(word, model) * np_max(model.word_weights) -# return 1 -# else: -# return 0 # Word not taken into account - + pass \ No newline at end of file diff --git a/fse/models/pooling.py b/fse/models/pooling.py index b98670a..1fdc387 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -34,9 +34,9 @@ from __future__ import division from fse.models.base_s2v import BaseSentence2VecModel +from fse.models.utils import get_ft_word_vector from gensim.models.keyedvectors import BaseKeyedVectors -from gensim.models.utils_any2vec import ft_ngram_hashes from numpy import ( ndarray, @@ -46,6 +46,7 @@ zeros, amax as np_amax, maximum as np_maximum, + max as np_max, ) from typing import List @@ -55,197 +56,273 @@ logger = logging.getLogger(__name__) +# def train_pooling_np( +# model: BaseSentence2VecModel, +# indexed_sentences: List[tuple], +# target: ndarray, +# memory: tuple, +# ) -> [int, int]: +# """Training on a sequence of sentences and update the target ndarray. + +# Called internally from :meth:`~fse.models.pooling.MaxPooling._do_train_job`. + +# Warnings +# -------- +# This is the non-optimized, pure Python version. If you have a C compiler, +# fse will use an optimized code path from :mod:`fse.models.pooling_inner` instead. + +# Parameters +# ---------- +# model : :class:`~fse.models.base_s2v.BaseSentence2VecModel` +# The BaseSentence2VecModel model instance. +# indexed_sentences : iterable of tuple +# The sentences used to train the model. +# target : ndarray +# The target ndarray. We use the index from indexed_sentences +# to write into the corresponding row of target. +# memory : tuple +# Private memory array(s) for each working thread + +# Returns +# ------- +# int, int +# Number of effective sentences (non-zero) and effective words in the vocabulary used +# during training the sentence embedding. + +# """ +# size = model.wv.vector_size +# vocab = model.wv.vocab + +# w_vectors = model.wv.vectors +# w_weights = model.word_weights + +# s_vectors = target + +# is_ft = model.is_ft + +# mem = memory[0] + +# hierarchical = model.hierarchical +# window_size = model.window_size +# window_stride = model.window_stride + +# if is_ft: +# # NOTE: For Fasttext: Use wv.vectors_vocab +# # Using the wv.vectors from fasttext had horrible effects on the sts results +# # I suspect this is because the wv.vectors are based on the averages of +# # wv.vectors_vocab + wv.vectors_ngrams, which will all point into very +# # similar directions. +# max_ngrams = model.batch_ngrams +# w_vectors = model.wv.vectors_vocab +# ngram_vectors = model.wv.vectors_ngrams +# min_n = model.wv.min_n +# max_n = model.wv.max_n +# bucket = model.wv.bucket +# oov_weight = np_amax(w_weights) + +# def get_ft_vector(word: str) -> ndarray: +# """ Function to compute the FT vectors if applicable + +# Parameters +# ---------- +# word : str +# String representation of token + +# Returns +# ------- +# ndarray +# FT vector representation +# """ +# if word in vocab: +# vocab_index = vocab[word].index +# return w_vectors[vocab_index] * w_weights[vocab_index] +# else: +# # Requires additional temporary storage +# ngram_hashes = ft_ngram_hashes(word, min_n, max_n, bucket, True)[ +# :max_ngrams +# ] +# if len(ngram_hashes) == 0: +# return zeros(size, dtype=REAL) +# return ( +# oov_weight +# * np_sum(ngram_vectors[ngram_hashes], axis=0) +# / len(ngram_hashes) +# ) + +# eff_sentences, eff_words = 0, 0 + +# if not is_ft: +# for obj in indexed_sentences: +# mem.fill(0.0) +# sent = obj[0] +# sent_adr = obj[1] + +# word_indices = [ +# vocab[word].index for word in sent if word in vocab +# ] +# eff_sentences += 1 +# if not len(word_indices): +# continue +# eff_words += len(word_indices) + +# if not hierarchical: +# # Take the maxium value along the axis +# mem = np_amax( +# np_mult( +# w_vectors[word_indices], +# w_weights[word_indices][:, None] +# ), +# axis=0, +# ) +# s_vectors[sent_adr] = np_maximum(s_vectors[sent_adr], mem,) +# else: +# # More expensive iteration +# for word_index, _ in enumerate(word_indices): +# if word_index % window_stride != 0: +# continue +# # Compute the local window +# window_indices = word_indices[word_index : word_index + window_size] +# # Perform average pooling [0,1,2,3,4] +# mem = np_sum( +# np_mult( +# w_vectors[window_indices], +# w_weights[window_indices][:, None], +# ), +# axis=0, +# ) +# # Perform hierarchical max pooling +# mem /= len(window_indices) +# s_vectors[sent_adr] = np_maximum(s_vectors[sent_adr], mem,) +# else: +# for obj in indexed_sentences: +# mem.fill(0.0) +# sent = obj[0] +# sent_adr = obj[1] + +# if not len(sent): +# continue +# mem = zeros(size, dtype=REAL) + +# eff_sentences += 1 +# eff_words += len(sent) # Counts everything in the sentence + +# if not hierarchical: +# for word in sent: +# s_vectors[sent_adr] = np_maximum( +# get_ft_vector(word), s_vectors[sent_adr], +# ) +# else: +# # if sent_adr == 4: +# # print("hi") + +# for word_index, word in enumerate(sent): +# if word_index % window_stride != 0: +# continue +# mem.fill(0.0) +# mem += get_ft_vector(word) +# count = 1 + +# for context in sent[word_index : word_index + window_size]: +# if word == context: +# continue +# mem += get_ft_vector(context) +# count += 1 +# mem /= count +# s_vectors[sent_adr] = np_maximum(mem, s_vectors[sent_adr],) + +# return eff_sentences, eff_words + +from fse.models.base_iterator import base_iterator, sentence_length + +def average_window_kernel( + model, + word : str, + mem : ndarray, + mem2 : ndarray, + ) -> int: + """ Window kernel implements aggregation function for window. + Does the vector conversion. + All results will be stored in mem. + """ + if word in model.wv.vocab: + word_index = model.wv.vocab[word].index + mem += model.wv.vectors[word_index] * model.word_weights[word_index] + return 1 + else: + if model.is_ft: + mem += get_ft_word_vector(word, model) * np_max(model.word_weights) + return 1 + else: + return 0 # Word not taken into account + return 1 + +def average_window_scaler( + model, + window_length : int, + mem : ndarray, + mem2 : ndarray, + ) -> None: + """ Window scaler implements scaling function for window result. + All results will be stored in mem2. + """ + mem /= window_length + +def max_sentence_kernel( + sent_length : int, + sent_index : int, + target : ndarray, + mem : ndarray, + mem2 : ndarray, + ) -> int: + """ Sentence kernel implements aggregation function for all windows. + All results will be stored in target. + """ + target[sent_index] = np_maximum(mem, target[sent_index],) + +def max_sentence_scaler( + win_count : int, + sent_adr : int, + target : ndarray, + mem : ndarray, + mem2 : ndarray, + ) -> None: + """ Sentence scaler implements scaling function for accumulated window result. + All results will be stored in target. + """ + # Does nothing, because max pooling + pass + def train_pooling_np( model: BaseSentence2VecModel, indexed_sentences: List[tuple], target: ndarray, memory: tuple, ) -> [int, int]: - """Training on a sequence of sentences and update the target ndarray. - - Called internally from :meth:`~fse.models.pooling.MaxPooling._do_train_job`. - - Warnings - -------- - This is the non-optimized, pure Python version. If you have a C compiler, - fse will use an optimized code path from :mod:`fse.models.pooling_inner` instead. - - Parameters - ---------- - model : :class:`~fse.models.base_s2v.BaseSentence2VecModel` - The BaseSentence2VecModel model instance. - indexed_sentences : iterable of tuple - The sentences used to train the model. - target : ndarray - The target ndarray. We use the index from indexed_sentences - to write into the corresponding row of target. - memory : tuple - Private memory array(s) for each working thread - - Returns - ------- - int, int - Number of effective sentences (non-zero) and effective words in the vocabulary used - during training the sentence embedding. - - """ - size = model.wv.vector_size - vocab = model.wv.vocab - - w_vectors = model.wv.vectors - w_weights = model.word_weights - - s_vectors = target - - is_ft = model.is_ft - - mem = memory[0] - - hierarchical = model.hierarchical - window_size = model.window_size - window_stride = model.window_stride - - if is_ft: - # NOTE: For Fasttext: Use wv.vectors_vocab - # Using the wv.vectors from fasttext had horrible effects on the sts results - # I suspect this is because the wv.vectors are based on the averages of - # wv.vectors_vocab + wv.vectors_ngrams, which will all point into very - # similar directions. - max_ngrams = model.batch_ngrams - w_vectors = model.wv.vectors_vocab - ngram_vectors = model.wv.vectors_ngrams - min_n = model.wv.min_n - max_n = model.wv.max_n - bucket = model.wv.bucket - oov_weight = np_amax(w_weights) - - def get_ft_vector(word: str) -> ndarray: - """ Function to compute the FT vectors if applicable - - Parameters - ---------- - word : str - String representation of token - - Returns - ------- - ndarray - FT vector representation - - """ - if word in vocab: - vocab_index = vocab[word].index - return w_vectors[vocab_index] * w_weights[vocab_index] - else: - # Requires additional temporary storage - ngram_hashes = ft_ngram_hashes(word, min_n, max_n, bucket, True)[ - :max_ngrams - ] - if len(ngram_hashes) == 0: - return zeros(size, dtype=REAL) - return ( - oov_weight - * np_sum(ngram_vectors[ngram_hashes], axis=0) - / len(ngram_hashes) - ) - - eff_sentences, eff_words = 0, 0 - - if not is_ft: - for obj in indexed_sentences: - mem.fill(0.0) - sent = obj[0] - sent_adr = obj[1] - - word_indices = [ - vocab[word].index for word in sent if word in vocab - ] - eff_sentences += 1 - if not len(word_indices): - continue - eff_words += len(word_indices) - - if not hierarchical: - # Take the maxium value along the axis - mem = np_amax( - np_mult( - w_vectors[word_indices], - w_weights[word_indices][:, None] - ), - axis=0, - ) - s_vectors[sent_adr] = np_maximum(s_vectors[sent_adr], mem,) - else: - # More expensive iteration - for word_index, _ in enumerate(word_indices): - if word_index % window_stride != 0: - continue - # Compute the local window - window_indices = word_indices[word_index : word_index + window_size] - # Perform average pooling [0,1,2,3,4] - mem = np_sum( - np_mult( - w_vectors[window_indices], - w_weights[window_indices][:, None], - ), - axis=0, - ) - # Perform hierarchical max pooling - mem /= len(window_indices) - s_vectors[sent_adr] = np_maximum(s_vectors[sent_adr], mem,) - else: - for obj in indexed_sentences: - mem.fill(0.0) - sent = obj[0] - sent_adr = obj[1] - - if not len(sent): - continue - mem = zeros(size, dtype=REAL) - - eff_sentences += 1 - eff_words += len(sent) # Counts everything in the sentence - - if not hierarchical: - for word in sent: - s_vectors[sent_adr] = np_maximum( - get_ft_vector(word), s_vectors[sent_adr], - ) - else: - if sent_adr == 4: - print("hi") - - for word_index, word in enumerate(sent): - if word_index % window_stride != 0: - continue - mem.fill(0.0) - mem += get_ft_vector(word) - count = 1 - - for context in sent[word_index : word_index + window_size]: - if word == context: - continue - mem += get_ft_vector(context) - count += 1 - mem /= count - s_vectors[sent_adr] = np_maximum(mem, s_vectors[sent_adr],) - - return eff_sentences, eff_words - - -try: - from fse.models.pooling_inner import train_pooling_cy - from fse.models.pooling_inner import ( - FAST_VERSION, - MAX_WORDS_IN_BATCH, - MAX_NGRAMS_IN_BATCH, + return base_iterator( + model=model, + indexed_sentences=indexed_sentences, + target=target, + memory=memory, + sentence_length=sentence_length, + window_kernel=average_window_kernel, + window_scaler=average_window_scaler, + sentence_kernel=max_sentence_kernel, + sentence_scaler=max_sentence_scaler, ) - train_pooling = train_pooling_cy -except ImportError: - FAST_VERSION = -1 - MAX_WORDS_IN_BATCH = 10000 - MAX_NGRAMS_IN_BATCH = 40 - train_pooling = train_pooling_np + +# try: +# from fse.models.pooling_inner import train_pooling_cy +# from fse.models.pooling_inner import ( +# FAST_VERSION, +# MAX_WORDS_IN_BATCH, +# MAX_NGRAMS_IN_BATCH, +# ) +# train_pooling = train_pooling_cy +# except ImportError: +FAST_VERSION = -1 +MAX_WORDS_IN_BATCH = 10000 +MAX_NGRAMS_IN_BATCH = 40 +train_pooling = train_pooling_np class MaxPooling(BaseSentence2VecModel): @@ -275,8 +352,7 @@ class MaxPooling(BaseSentence2VecModel): def __init__( self, model: BaseKeyedVectors, - hierarchical: bool = False, - window_size: int = 5, + window_size: int = 1, window_stride: int = 1, sv_mapfile_path: str = None, wv_mapfile_path: str = None, @@ -308,7 +384,6 @@ def __init__( Number of working threads, used for multithreading. For most tasks (few words in a sentence) a value of 1 should be more than enough. """ - self.hierarchical = bool(hierarchical) self.window_size = int(window_size) self.window_stride = int(window_stride) diff --git a/fse/test/test_pooling.py b/fse/test/test_pooling.py index 869f794..32218c7 100644 --- a/fse/test/test_pooling.py +++ b/fse/test/test_pooling.py @@ -5,6 +5,53 @@ # Copyright (C) 2020 Oliver Borchers # For License information, see corresponding LICENSE file. +""" +Example for computation of convolution length + +window=5 +stride=3 + +window=5 +stride=1 + +Consider, that w2v does not contain "12345" + +all w2v ft +"They", "admit" +2 2 2 + 1 1 + +"So", "Apple", "bought", "buds" +4 4 4 +1 3 3 + 2 2 + 1 1 + +"go", "12345" +2 1 2 + 1 + +"pull", "12345678910111213" +2 0 2 + 1 + +"this" "is" "a" "longer" "test" "sentence" "test" "longer" "sentences" +0 1 2 3 4 + 0 1 2 3 4 + 0 1 2 + +"this" 5 5 5 +"is" 4 5 +"a" 4 5 +"longer" 5 4 5 +"test" 3 5 +"sentence" 2 4 +"test" 3 2 3 +"longer" 1 2 +"sentences" 0 1 +""" + + from fse.models.pooling import MaxPooling, train_pooling_np from fse.test.model_shared_imports import * @@ -25,6 +72,14 @@ def setUp(self): ) self.model._pre_train_calls() + def set_convolution(self): + self.model.window_size=5 + self.model.window_stride=1 + + def unset_convolution(self): + self.model.window_size=1 + self.model.window_stride=1 + @unittest.skipIf(IGNORE_CY, "ignoring Cython build") def test_cython(self): from fse.models.pooling_inner import ( @@ -239,41 +294,43 @@ def test_pool_np_ft_non_negative(self): ## Hierarchical Tests start here - def test_hier_pool_train_np_w2v(self): + def test_conv_pool_train_np_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() - self.model.hierarchical = True + self.set_convolution() output = train_pooling_np( self.model, self.sentences, self.model.sv.vectors, mem ) - self.model.hierarchical = False + self.unset_convolution() - self.assertEqual((5, 14), output) + # TODO: The count does not match the expectation + self.assertEqual((5, 39), output) self.assertTrue((183 == self.model.sv[0]).all()) - self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) + self.assertTrue(np.allclose(self.model.sv[4], 184.8)) @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_hier_pool_train_cy_w2v(self): + def test_conv_pool_train_cy_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() - self.model.hierarchical = True + self.set_convolution() from fse.models.pooling_inner import train_pooling_cy output = train_pooling_cy( self.model, self.sentences, self.model.sv.vectors, mem ) - self.model.hierarchical = False - self.assertEqual((5, 14), output) + self.unset_convolution() + + self.assertEqual((5, 39), output) self.assertTrue((183 == self.model.sv[0]).all()) - self.assertTrue(np.allclose(self.model.sv[4], 245.66667)) + self.assertTrue(np.allclose(self.model.sv[4], 184.8)) - def test_hier_pool_train_np_ft(self): - m = MaxPooling(FT_DET, hierarchical=True) + def test_conv_pool_train_np_ft(self): + m = MaxPooling(FT_DET, window_size=2, window_stride=2) m.prep.prepare_vectors( sv=m.sv, total_sentences=len(self.sentences), update=False ) @@ -284,8 +341,8 @@ def test_hier_pool_train_np_ft(self): self.assertEqual((5, 19), output) self.assertTrue((1 == m.sv[0]).all()) - self.assertTrue(np.allclose(737413.9, m.sv[2])) - self.assertTrue(np.allclose(1080970.2, m.sv[3])) + self.assertTrue(np.allclose(368707.44, m.sv[2])) + self.assertTrue(np.allclose(961940.2, m.sv[3])) """ Note to future self: Due to the size of the ngram vectors, @@ -295,8 +352,8 @@ def test_hier_pool_train_np_ft(self): """ @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_hier_pool_train_cy_ft(self): - m = MaxPooling(FT_DET, hierarchical=True) + def test_conv_pool_train_cy_ft(self): + m = MaxPooling(FT_DET, window_size=2, window_stride=2) m.prep.prepare_vectors( sv=m.sv, total_sentences=len(self.sentences), update=False ) @@ -309,12 +366,12 @@ def test_hier_pool_train_cy_ft(self): self.assertEqual((5, 19), output) self.assertTrue((1 == m.sv[0]).all()) - self.assertTrue(np.allclose(737413.9, m.sv[2])) - self.assertTrue(np.allclose(1080970.2, m.sv[3])) + self.assertTrue(np.allclose(368707.44, m.sv[2])) + self.assertTrue(np.allclose(961940.2, m.sv[3])) @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_hier_pool_cy_equal_np_w2v_random(self): - m1 = MaxPooling(W2V_RNG, hierarchical=True) + def test_conv_pool_cy_equal_np_w2v_random(self): + m1 = MaxPooling(W2V_RNG, window_size=5, window_stride=1) m1.prep.prepare_vectors( sv=m1.sv, total_sentences=len(self.sentences), update=False ) @@ -322,7 +379,7 @@ def test_hier_pool_cy_equal_np_w2v_random(self): mem1 = m1._get_thread_working_mem() o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) - m2 = MaxPooling(W2V_RNG, hierarchical=True) + m2 = MaxPooling(W2V_RNG, window_size=5, window_stride=1) m2.prep.prepare_vectors( sv=m2.sv, total_sentences=len(self.sentences), update=False ) @@ -336,8 +393,8 @@ def test_hier_pool_cy_equal_np_w2v_random(self): self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_hier_pool_cy_equal_np_ft_random(self): - m1 = MaxPooling(FT_RNG, hierarchical=True) + def test_conv_pool_cy_equal_np_ft_random(self): + m1 = MaxPooling(FT_RNG, window_size=5, window_stride=1) m1.prep.prepare_vectors( sv=m1.sv, total_sentences=len(self.sentences), update=False ) @@ -349,7 +406,7 @@ def test_hier_pool_cy_equal_np_ft_random(self): mem1 = m1._get_thread_working_mem() o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - m2 = MaxPooling(FT_RNG, hierarchical=True) + m2 = MaxPooling(FT_RNG, window_size=5, window_stride=1) m2.prep.prepare_vectors( sv=m2.sv, total_sentences=len(self.sentences), update=False ) @@ -364,42 +421,41 @@ def test_hier_pool_cy_equal_np_ft_random(self): self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_hier_pool_cy_w2v_non_negative(self): - mpool = MaxPooling(W2V_RNG, hierarchical=True) + def test_conv_pool_cy_w2v_non_negative(self): + mpool = MaxPooling(W2V_RNG, window_size=5, window_stride=1) mpool.train(self.sentences) self.assertTrue((mpool.sv.vectors >= 0).all()) @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_hier_pool_cy_ft_non_negative(self): - mpool = MaxPooling(FT_RNG, hierarchical=True) + def test_conv_pool_cy_ft_non_negative(self): + mpool = MaxPooling(FT_RNG, window_size=5, window_stride=1) mpool.train(self.sentences) self.assertTrue((mpool.sv.vectors >= 0).all()) ### Hierarchical Test + Stride start here - def test_hier_pool_stride_train_np_w2v(self): + def test_conv_pool_stride_train_np_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() - self.model.hierarchical = True + self.set_convolution() self.model.window_stride = 5 output = train_pooling_np( self.model, self.sentences, self.model.sv.vectors, mem ) - self.model.hierarchical = False - self.model.window_stride = 1 + self.unset_convolution() self.assertEqual((5, 14), output) self.assertTrue((183 == self.model.sv[0]).all()) - self.assertTrue((231 == self.model.sv[4]).all()) + self.assertTrue((115.5 == self.model.sv[4]).all()) @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_hier_pool_stride_train_cy_w2v(self): + def test_conv_pool_stride_train_cy_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() - self.model.hierarchical = True + self.set_convolution() self.model.window_stride = 5 from fse.models.pooling_inner import train_pooling_cy @@ -407,15 +463,14 @@ def test_hier_pool_stride_train_cy_w2v(self): output = train_pooling_cy( self.model, self.sentences, self.model.sv.vectors, mem ) - self.model.hierarchical = False - self.model.window_stride = 1 + self.unset_convolution() self.assertEqual((5, 14), output) self.assertTrue((183 == self.model.sv[0]).all()) - self.assertTrue((231 == self.model.sv[4]).all()) + self.assertTrue((115.5 == self.model.sv[4]).all()) - def test_hier_pool_stride_train_np_ft(self): - m = MaxPooling(FT_DET, hierarchical=True, window_stride=3) + def test_conv_pool_stride_train_np_ft(self): + m = MaxPooling(FT_DET, window_size=5, window_stride=3) m.prep.prepare_vectors( sv=m.sv, total_sentences=len(self.sentences), update=False ) @@ -424,14 +479,14 @@ def test_hier_pool_stride_train_np_ft(self): output = train_pooling_np(m, self.sentences, m.sv.vectors, mem) - self.assertEqual((5, 19), output) + self.assertEqual((5, 24), output) self.assertTrue((1 == m.sv[0]).all()) self.assertTrue(np.allclose(368707.44, m.sv[2])) self.assertTrue(np.allclose(961940.2, m.sv[3])) @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_hier_pool_stride_train_cy_ft(self): - m = MaxPooling(FT_DET, hierarchical=True, window_stride=3) + def test_conv_pool_stride_train_cy_ft(self): + m = MaxPooling(FT_DET, window_size=5, window_stride=3) m.prep.prepare_vectors( sv=m.sv, total_sentences=len(self.sentences), update=False ) @@ -442,14 +497,14 @@ def test_hier_pool_stride_train_cy_ft(self): output = train_pooling_cy(m, self.sentences, m.sv.vectors, mem) - self.assertEqual((5, 19), output) + self.assertEqual((5, 24), output) self.assertTrue((1 == m.sv[0]).all()) self.assertTrue(np.allclose(368707.44, m.sv[2])) self.assertTrue(np.allclose(961940.2, m.sv[3])) @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_hier_pool_stride_cy_equal_np_w2v_random(self): - m1 = MaxPooling(W2V_RNG, hierarchical=True, window_stride=4) + def test_conv_pool_stride_cy_equal_np_w2v_random(self): + m1 = MaxPooling(W2V_RNG, window_size=5, window_stride=4) m1.prep.prepare_vectors( sv=m1.sv, total_sentences=len(self.sentences), update=False ) @@ -457,7 +512,7 @@ def test_hier_pool_stride_cy_equal_np_w2v_random(self): mem1 = m1._get_thread_working_mem() o1 = train_pooling_np(m1, self.sentences, m1.sv.vectors, mem1) - m2 = MaxPooling(W2V_RNG, hierarchical=True, window_stride=4) + m2 = MaxPooling(W2V_RNG, window_size=5, window_stride=4) m2.prep.prepare_vectors( sv=m2.sv, total_sentences=len(self.sentences), update=False ) @@ -471,8 +526,8 @@ def test_hier_pool_stride_cy_equal_np_w2v_random(self): self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_hier_pool_stride_cy_equal_np_ft_random(self): - m1 = MaxPooling(FT_RNG, hierarchical=True, window_stride=5) + def test_conv_pool_stride_cy_equal_np_ft_random(self): + m1 = MaxPooling(FT_RNG, window_size=5, window_stride=5) m1.prep.prepare_vectors( sv=m1.sv, total_sentences=len(self.sentences), update=False ) @@ -484,7 +539,7 @@ def test_hier_pool_stride_cy_equal_np_ft_random(self): mem1 = m1._get_thread_working_mem() o1 = train_pooling_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - m2 = MaxPooling(FT_RNG, hierarchical=True, window_stride=5) + m2 = MaxPooling(FT_RNG, window_size=5, window_stride=5) m2.prep.prepare_vectors( sv=m2.sv, total_sentences=len(self.sentences), update=False ) @@ -498,13 +553,13 @@ def test_hier_pool_stride_cy_equal_np_ft_random(self): self.assertEqual(o1, o2) self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - def test_hier_pool_stride_np_w2v_non_negative(self): - mpool = MaxPooling(W2V_RNG, hierarchical=True, window_stride=4) + def test_conv_pool_stride_np_w2v_non_negative(self): + mpool = MaxPooling(W2V_RNG, window_size=2, window_stride=2) mpool.train(self.sentences) self.assertTrue((mpool.sv.vectors >= 0).all()) - def test_hier_pool_stride_np_ft_non_negative(self): - mpool = MaxPooling(FT_RNG, hierarchical=True, window_stride=4) + def test_conv_pool_stride_np_ft_non_negative(self): + mpool = MaxPooling(FT_RNG, window_size=2, window_stride=2) mpool.train(self.sentences) self.assertTrue((mpool.sv.vectors >= 0).all()) From bc0e668801c47fa1dd563d3492e95203874a2f4c Mon Sep 17 00:00:00 2001 From: OliverB Date: Thu, 6 Aug 2020 18:47:17 +0200 Subject: [PATCH 62/63] changed readme --- README.md | 1 + 1 file changed, 1 insertion(+) diff --git a/README.md b/README.md index 812b422..5acf906 100644 --- a/README.md +++ b/README.md @@ -14,6 +14,7 @@ Fast Sentence Embeddings is a Python library that serves as an addition to Gensi If you want to support fse, take a quick [survey](https://forms.gle/8uSU323fWUVtVwcAA) to improve it :-) +*For additional features, please check out the dev branch. There is no development on the master branch* Features ------------ From ac1745165a7042b3a4f85cccc673b3f33581ad8c Mon Sep 17 00:00:00 2001 From: OliverB Date: Wed, 17 Mar 2021 17:37:10 +0100 Subject: [PATCH 63/63] Last changes --- .gitignore | 1 + .gitignore.save | 61 ++++++++++ fse/models/average.py | 6 +- fse/models/pooling.py | 180 +-------------------------- fse/test/test_average.py | 257 ++++++++++++++------------------------- 5 files changed, 159 insertions(+), 346 deletions(-) create mode 100644 .gitignore.save diff --git a/.gitignore b/.gitignore index 3433c1c..2c7da37 100644 --- a/.gitignore +++ b/.gitignore @@ -46,6 +46,7 @@ Thumbs.db legacy latex draft +drafts fse.egg-info/ # Other # diff --git a/.gitignore.save b/.gitignore.save new file mode 100644 index 0000000..1ab2928 --- /dev/null +++ b/.gitignore.save @@ -0,0 +1,61 @@ +# Compiled source # +################### +*.com +*.class +*.dll +*.exe +*.o +*.so +*.pyc + +# Packages # +############ +# it's better to unpack these files and commit the raw source +# git has its own built in compression methods +*.7z +*.dmg +*.gz +*.iso +*.jar +*.rar +*.tar +*.zip + +# Logs and databases # +###################### +*.log +*.sql +*.sqlite +*.pkl +*.bak +*.npy +*.npz +*.code-workspace + +# OS generated files # +###################### +.DS_Store? +.DS_Store +ehthumbs.db +Icon? +Thumbs.db +*.icloud + +# Folders # +########### +legacy + +# Other # +######### +.ipynb_checkpoints/ +.settings/ +.vscode/ +.eggs +.coverage +*.bak +/build/ +/dist/ +*.prof +*.lprof +*.bin +*.old diff --git a/fse/models/average.py b/fse/models/average.py index 52dc8f7..1feddfe 100644 --- a/fse/models/average.py +++ b/fse/models/average.py @@ -209,6 +209,8 @@ class Average(BaseSentence2VecModel): def __init__( self, model: BaseKeyedVectors, + window_size: int = 1, + window_stride: int = 1, sv_mapfile_path: str = None, wv_mapfile_path: str = None, workers: int = 1, @@ -242,7 +244,9 @@ def __init__( See https://github.com/LuminosoInsight/wordfreq """ - + self.window_size = int(window_size) + self.window_stride = int(window_stride) + super(Average, self).__init__( model=model, sv_mapfile_path=sv_mapfile_path, diff --git a/fse/models/pooling.py b/fse/models/pooling.py index 1fdc387..525c098 100644 --- a/fse/models/pooling.py +++ b/fse/models/pooling.py @@ -35,6 +35,7 @@ from fse.models.base_s2v import BaseSentence2VecModel from fse.models.utils import get_ft_word_vector +from fse.models.base_iterator import base_iterator, sentence_length from gensim.models.keyedvectors import BaseKeyedVectors @@ -55,185 +56,6 @@ logger = logging.getLogger(__name__) - -# def train_pooling_np( -# model: BaseSentence2VecModel, -# indexed_sentences: List[tuple], -# target: ndarray, -# memory: tuple, -# ) -> [int, int]: -# """Training on a sequence of sentences and update the target ndarray. - -# Called internally from :meth:`~fse.models.pooling.MaxPooling._do_train_job`. - -# Warnings -# -------- -# This is the non-optimized, pure Python version. If you have a C compiler, -# fse will use an optimized code path from :mod:`fse.models.pooling_inner` instead. - -# Parameters -# ---------- -# model : :class:`~fse.models.base_s2v.BaseSentence2VecModel` -# The BaseSentence2VecModel model instance. -# indexed_sentences : iterable of tuple -# The sentences used to train the model. -# target : ndarray -# The target ndarray. We use the index from indexed_sentences -# to write into the corresponding row of target. -# memory : tuple -# Private memory array(s) for each working thread - -# Returns -# ------- -# int, int -# Number of effective sentences (non-zero) and effective words in the vocabulary used -# during training the sentence embedding. - -# """ -# size = model.wv.vector_size -# vocab = model.wv.vocab - -# w_vectors = model.wv.vectors -# w_weights = model.word_weights - -# s_vectors = target - -# is_ft = model.is_ft - -# mem = memory[0] - -# hierarchical = model.hierarchical -# window_size = model.window_size -# window_stride = model.window_stride - -# if is_ft: -# # NOTE: For Fasttext: Use wv.vectors_vocab -# # Using the wv.vectors from fasttext had horrible effects on the sts results -# # I suspect this is because the wv.vectors are based on the averages of -# # wv.vectors_vocab + wv.vectors_ngrams, which will all point into very -# # similar directions. -# max_ngrams = model.batch_ngrams -# w_vectors = model.wv.vectors_vocab -# ngram_vectors = model.wv.vectors_ngrams -# min_n = model.wv.min_n -# max_n = model.wv.max_n -# bucket = model.wv.bucket -# oov_weight = np_amax(w_weights) - -# def get_ft_vector(word: str) -> ndarray: -# """ Function to compute the FT vectors if applicable - -# Parameters -# ---------- -# word : str -# String representation of token - -# Returns -# ------- -# ndarray -# FT vector representation -# """ -# if word in vocab: -# vocab_index = vocab[word].index -# return w_vectors[vocab_index] * w_weights[vocab_index] -# else: -# # Requires additional temporary storage -# ngram_hashes = ft_ngram_hashes(word, min_n, max_n, bucket, True)[ -# :max_ngrams -# ] -# if len(ngram_hashes) == 0: -# return zeros(size, dtype=REAL) -# return ( -# oov_weight -# * np_sum(ngram_vectors[ngram_hashes], axis=0) -# / len(ngram_hashes) -# ) - -# eff_sentences, eff_words = 0, 0 - -# if not is_ft: -# for obj in indexed_sentences: -# mem.fill(0.0) -# sent = obj[0] -# sent_adr = obj[1] - -# word_indices = [ -# vocab[word].index for word in sent if word in vocab -# ] -# eff_sentences += 1 -# if not len(word_indices): -# continue -# eff_words += len(word_indices) - -# if not hierarchical: -# # Take the maxium value along the axis -# mem = np_amax( -# np_mult( -# w_vectors[word_indices], -# w_weights[word_indices][:, None] -# ), -# axis=0, -# ) -# s_vectors[sent_adr] = np_maximum(s_vectors[sent_adr], mem,) -# else: -# # More expensive iteration -# for word_index, _ in enumerate(word_indices): -# if word_index % window_stride != 0: -# continue -# # Compute the local window -# window_indices = word_indices[word_index : word_index + window_size] -# # Perform average pooling [0,1,2,3,4] -# mem = np_sum( -# np_mult( -# w_vectors[window_indices], -# w_weights[window_indices][:, None], -# ), -# axis=0, -# ) -# # Perform hierarchical max pooling -# mem /= len(window_indices) -# s_vectors[sent_adr] = np_maximum(s_vectors[sent_adr], mem,) -# else: -# for obj in indexed_sentences: -# mem.fill(0.0) -# sent = obj[0] -# sent_adr = obj[1] - -# if not len(sent): -# continue -# mem = zeros(size, dtype=REAL) - -# eff_sentences += 1 -# eff_words += len(sent) # Counts everything in the sentence - -# if not hierarchical: -# for word in sent: -# s_vectors[sent_adr] = np_maximum( -# get_ft_vector(word), s_vectors[sent_adr], -# ) -# else: -# # if sent_adr == 4: -# # print("hi") - -# for word_index, word in enumerate(sent): -# if word_index % window_stride != 0: -# continue -# mem.fill(0.0) -# mem += get_ft_vector(word) -# count = 1 - -# for context in sent[word_index : word_index + window_size]: -# if word == context: -# continue -# mem += get_ft_vector(context) -# count += 1 -# mem /= count -# s_vectors[sent_adr] = np_maximum(mem, s_vectors[sent_adr],) - -# return eff_sentences, eff_words - -from fse.models.base_iterator import base_iterator, sentence_length - def average_window_kernel( model, word : str, diff --git a/fse/test/test_average.py b/fse/test/test_average.py index d5bd454..e72ca56 100644 --- a/fse/test/test_average.py +++ b/fse/test/test_average.py @@ -1,213 +1,159 @@ #!/usr/bin/env python # -*- coding: utf-8 -*- -# Author: Oliver Borchers -# Copyright (C) 2020 Oliver Borchers -# For License information, see corresponding LICENSE file. +# Author: Oliver Borchers +# Copyright (C) 2019 Oliver Borchers + """ Automated tests for checking the average model. """ -from fse.test.model_shared_imports import * +import logging +import unittest + +from pathlib import Path + +import numpy as np + +from fse.models.average import Average +from fse.models.average import train_average_np +from fse.models.average_inner import train_average_cy +from fse.models.average_inner import FAST_VERSION, MAX_WORDS_IN_BATCH +from fse.inputs import IndexedSentence + +from gensim.models import Word2Vec, FastText + +logger = logging.getLogger(__name__) + +CORPUS = Path("fse/test/test_data/test_sentences.txt") +DIM = 5 +W2V = Word2Vec(min_count=1, size=DIM) +SENTENCES = [l.split() for i, l in enumerate(open(CORPUS, "r"))] +W2V.build_vocab(SENTENCES) +W2V.wv.vectors[:,] = np.arange(len(W2V.wv.vectors), dtype=np.float32)[:, None] -from fse.models.average import Average, train_average_np class TestAverageFunctions(unittest.TestCase): def setUp(self): - self.sentences = [ - ["They", "admit"], - ["So", "Apple", "bought", "buds"], - ["go", "12345"], - ["pull", "12345678910111213"], - ] - self.sentences = [(s, i) for i, s in enumerate(self.sentences)] - self.model = Average(W2V_DET) - self.model.prep.prepare_vectors( - sv=self.model.sv, total_sentences=len(self.sentences), update=False - ) + self.sentences = [["They", "admit"], ["So", "Apple", "bought", "buds"], ["go", "12345"], ["pull", "12345678910111213"]] + self.sentences = [IndexedSentence(s, i) for i,s in enumerate(self.sentences)] + self.model = Average(W2V) + self.model.prep.prepare_vectors(sv=self.model.sv, total_sentences=len(self.sentences), update=False) self.model._pre_train_calls() - @unittest.skipIf(IGNORE_CY, "ignoring Cython build") def test_cython(self): - from fse.models.average_inner import ( - FAST_VERSION, - MAX_WORDS_IN_BATCH, - MAX_NGRAMS_IN_BATCH, - ) - self.assertTrue(FAST_VERSION) - self.assertEqual(10000, MAX_WORDS_IN_BATCH) - self.assertEqual(40, MAX_NGRAMS_IN_BATCH) + self.assertEqual(10000,MAX_WORDS_IN_BATCH) - def test_average_train_np_w2v_det(self): + def test_average_train_np_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() - output = train_average_np( - self.model, self.sentences, self.model.sv.vectors, mem - ) + output = train_average_np(self.model, self.sentences, self.model.sv.vectors, mem) self.assertEqual((4, 7), output) self.assertTrue((183 == self.model.sv[0]).all()) self.assertTrue((164.5 == self.model.sv[1]).all()) self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) - @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_average_train_cy_w2v_det(self): + def test_average_train_cy_w2v(self): self.model.sv.vectors = np.zeros_like(self.model.sv.vectors, dtype=np.float32) mem = self.model._get_thread_working_mem() - - from fse.models.average_inner import train_average_cy - - output = train_average_cy( - self.model, self.sentences, self.model.sv.vectors, mem - ) + output = train_average_cy(self.model, self.sentences, self.model.sv.vectors, mem) self.assertEqual((4, 7), output) self.assertTrue((183 == self.model.sv[0]).all()) self.assertTrue((164.5 == self.model.sv[1]).all()) self.assertTrue((self.model.wv.vocab["go"].index == self.model.sv[2]).all()) - def test_average_train_np_ft_det(self): - m = Average(FT_DET) - m.prep.prepare_vectors( - sv=m.sv, total_sentences=len(self.sentences), update=False - ) + def test_average_train_np_ft(self): + ft = FastText(min_count=1, size=DIM) + ft.build_vocab(SENTENCES) + m = Average(ft) + m.prep.prepare_vectors(sv=m.sv, total_sentences=len(self.sentences), update=False) m._pre_train_calls() + m.wv.vectors = m.wv.vectors_vocab = np.ones_like(m.wv.vectors, dtype=np.float32) + m.wv.vectors_ngrams = np.full_like(m.wv.vectors_ngrams, 2, dtype=np.float32) mem = m._get_thread_working_mem() output = train_average_np(m, self.sentences, m.sv.vectors, mem) - self.assertEqual((4, 10), output) - self.assertTrue((1.0 + EPS == m.sv[0]).all()) - self.assertTrue(np.allclose(368707.44, m.sv[2])) - self.assertTrue(np.allclose(961940.2, m.sv[3])) + self.assertTrue((1. == m.sv[0]).all()) + self.assertTrue((1.5 == m.sv[2]).all()) + self.assertTrue((2 == m.sv[3]).all()) # "go" -> [1,1...] # oov: "12345" -> (14 hashes * 2) / 14 = 2 # (2 + 1) / 2 = 1.5 - @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_average_train_cy_ft_det(self): - m = Average(FT_DET) - m.prep.prepare_vectors( - sv=m.sv, total_sentences=len(self.sentences), update=False - ) + def test_average_train_cy_ft(self): + ft = FastText(min_count=1, size=DIM) + ft.build_vocab(SENTENCES) + m = Average(ft) + m.prep.prepare_vectors(sv=m.sv, total_sentences=len(self.sentences), update=False) m._pre_train_calls() + m.wv.vectors = m.wv.vectors_vocab = np.ones_like(m.wv.vectors, dtype=np.float32) + m.wv.vectors_ngrams = np.full_like(m.wv.vectors_ngrams, 2, dtype=np.float32) mem = m._get_thread_working_mem() - - from fse.models.average_inner import train_average_cy - output = train_average_cy(m, self.sentences, m.sv.vectors, mem) self.assertEqual((4, 10), output) - self.assertTrue((1.0 + EPS == m.sv[0]).all()) - self.assertTrue(np.allclose(368707.4, m.sv[2])) - self.assertTrue(np.allclose(961940.0, m.sv[3])) - - @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_cy_equal_np_w2v_det(self): - m1 = Average(W2V_DET) - m1.prep.prepare_vectors( - sv=m1.sv, total_sentences=len(self.sentences), update=False - ) + self.assertTrue((1. == m.sv[0]).all()) + self.assertTrue((1.5 == m.sv[2]).all()) + self.assertTrue((2 == m.sv[3]).all()) + + def test_cy_equal_np_w2v(self): + m1 = Average(W2V) + m1.prep.prepare_vectors(sv=m1.sv, total_sentences=len(self.sentences), update=False) m1._pre_train_calls() mem1 = m1._get_thread_working_mem() o1 = train_average_np(m1, self.sentences, m1.sv.vectors, mem1) - m2 = Average(W2V_DET) - m2.prep.prepare_vectors( - sv=m2.sv, total_sentences=len(self.sentences), update=False - ) + m2 = Average(W2V) + m2.prep.prepare_vectors(sv=m2.sv, total_sentences=len(self.sentences), update=False) m2._pre_train_calls() mem2 = m2._get_thread_working_mem() - - from fse.models.average_inner import train_average_cy - o2 = train_average_cy(m2, self.sentences, m2.sv.vectors, mem2) self.assertEqual(o1, o2) self.assertTrue((m1.sv.vectors == m2.sv.vectors).all()) - @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_cy_equal_np_w2v_rng(self): - m1 = Average(W2V_RNG) - m1.prep.prepare_vectors( - sv=m1.sv, total_sentences=len(self.sentences), update=False - ) + def test_cy_equal_np_ft(self): + ft = FastText(size=20, min_count=1) + ft.build_vocab(SENTENCES) + + m1 = Average(ft) + m1.prep.prepare_vectors(sv=m1.sv, total_sentences=len(self.sentences), update=False) m1._pre_train_calls() mem1 = m1._get_thread_working_mem() o1 = train_average_np(m1, self.sentences, m1.sv.vectors, mem1) - m2 = Average(W2V_RNG) - m2.prep.prepare_vectors( - sv=m2.sv, total_sentences=len(self.sentences), update=False - ) + m2 = Average(ft) + m2.prep.prepare_vectors(sv=m2.sv, total_sentences=len(self.sentences), update=False) m2._pre_train_calls() mem2 = m2._get_thread_working_mem() - - from fse.models.average_inner import train_average_cy - o2 = train_average_cy(m2, self.sentences, m2.sv.vectors, mem2) - self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) - - @unittest.skipIf(IGNORE_CY, "ignoring Cython build") - def test_cy_equal_np_ft_rng(self): - m1 = Average(FT_RNG) - m1.prep.prepare_vectors( - sv=m1.sv, total_sentences=len(self.sentences), update=False - ) - m1._pre_train_calls() - - from fse.models.average_inner import MAX_NGRAMS_IN_BATCH - - m1.batch_ngrams = MAX_NGRAMS_IN_BATCH - mem1 = m1._get_thread_working_mem() - o1 = train_average_np(m1, self.sentences[:2], m1.sv.vectors, mem1) - - m2 = Average(FT_RNG) - m2.prep.prepare_vectors( - sv=m2.sv, total_sentences=len(self.sentences), update=False - ) - m2._pre_train_calls() - mem2 = m2._get_thread_working_mem() - - from fse.models.average_inner import train_average_cy - - o2 = train_average_cy(m2, self.sentences[:2], m2.sv.vectors, mem2) - self.assertEqual(o1, o2) - self.assertTrue(np.allclose(m1.sv.vectors, m2.sv.vectors, atol=1e-6)) + self.assertTrue(np.allclose(m1.sv.vectors, )) + self.assertTrue(( == m2.sv.vectors).all()) def test_do_train_job(self): - self.model.prep.prepare_vectors( - sv=self.model.sv, total_sentences=len(SENTENCES), update=True - ) + self.model.prep.prepare_vectors(sv=self.model.sv, total_sentences=len(SENTENCES), update=True) mem = self.model._get_thread_working_mem() - self.assertEqual( - (100, 1450), - self.model._do_train_job( - [(s, i) for i, s in enumerate(SENTENCES)], - target=self.model.sv.vectors, - memory=mem, - ), + self.assertEqual((100,1450), self.model._do_train_job( + [IndexedSentence(s, i) for i,s in enumerate(SENTENCES)], + target=self.model.sv.vectors, memory=mem) ) - self.assertEqual((104, DIM), self.model.sv.vectors.shape) + self.assertEqual((104,DIM), self.model.sv.vectors.shape) def test_train(self): - self.assertEqual( - (100, 1450), self.model.train([(s, i) for i, s in enumerate(SENTENCES)]) - ) - + self.assertEqual((100,1450), self.model.train([IndexedSentence(s, i) for i,s in enumerate(SENTENCES)])) + def test_train_single_from_disk(self): p = Path("fse/test/test_data/test_vecs") p_res = Path("fse/test/test_data/test_vecs.vectors") p_target = Path("fse/test/test_data/test_vecs_wv.vectors") - se1 = Average(W2V_DET) - se2 = Average( - W2V_DET, - sv_mapfile_path=str(p.absolute()), - wv_mapfile_path=str(p.absolute()), - ) - se1.train([(s, i) for i, s in enumerate(SENTENCES)]) - se2.train([(s, i) for i, s in enumerate(SENTENCES)]) + se1 = Average(W2V) + se2 = Average(W2V, sv_mapfile_path=str(p.absolute()) ,wv_mapfile_path=str(p.absolute())) + se1.train([IndexedSentence(s, i) for i,s in enumerate(SENTENCES)]) + se2.train([IndexedSentence(s, i) for i,s in enumerate(SENTENCES)]) self.assertTrue(p_target.exists()) self.assertTrue((se1.wv.vectors == se2.wv.vectors).all()) @@ -222,15 +168,10 @@ def test_train_multi_from_disk(self): p_res = Path("fse/test/test_data/test_vecs.vectors") p_target = Path("fse/test/test_data/test_vecs_wv.vectors") - se1 = Average(W2V_DET, workers=2) - se2 = Average( - W2V_DET, - workers=2, - sv_mapfile_path=str(p.absolute()), - wv_mapfile_path=str(p.absolute()), - ) - se1.train([(s, i) for i, s in enumerate(SENTENCES)]) - se2.train([(s, i) for i, s in enumerate(SENTENCES)]) + se1 = Average(W2V, workers=2) + se2 = Average(W2V, workers=2, sv_mapfile_path=str(p.absolute()) ,wv_mapfile_path=str(p.absolute())) + se1.train([IndexedSentence(s, i) for i,s in enumerate(SENTENCES)]) + se2.train([IndexedSentence(s, i) for i,s in enumerate(SENTENCES)]) self.assertTrue(p_target.exists()) self.assertTrue((se1.wv.vectors == se2.wv.vectors).all()) @@ -241,27 +182,11 @@ def test_train_multi_from_disk(self): p_target.unlink() def test_check_parameter_sanity(self): - se = Average(W2V_DET) - se.word_weights = np.full(20, 2.0, dtype=np.float32) + se = Average(W2V) + se.word_weights = np.full(20, 2., dtype=np.float32) with self.assertRaises(ValueError): se._check_parameter_sanity() - # TODO: - # se = Average(W2V_DET, window_size=0) - # with self.assertRaises(ValueError): - # se._check_parameter_sanity() - - # se = Average(W2V_DET, window_size=3, window_stride=0) - # with self.assertRaises(ValueError): - # se._check_parameter_sanity() - - # se = Average(W2V_DET, window_size=3, window_stride=4) - # with self.assertRaises(ValueError): - # se._check_parameter_sanity() - - -if __name__ == "__main__": - logging.basicConfig( - format="%(asctime)s : %(levelname)s : %(message)s", level=logging.DEBUG - ) - unittest.main() +if __name__ == '__main__': + logging.basicConfig(format='%(asctime)s : %(levelname)s : %(message)s', level=logging.DEBUG) + unittest.main() \ No newline at end of file