Use lazy imports to speed up initial loading time

r2ai/interpreter.py

@@ -26,11 +26,9 @@
 from .models import get_hf_llm, new_get_hf_llm, get_default_model
 from .voice import tts
 from .const import R2AI_HOMEDIR
-from . import auto, LOGGER, logging
+from . import LOGGER, logging
 from .web import stop_http_server, server_running
 from .progress import progress_bar
-import litellm
-from .completion import messages_to_prompt
 
 file_dir = os.path.dirname(__file__)
 sys.path.append(file_dir)
@@ -90,6 +88,7 @@ def ddg(m):
     return f"Considering:\n```{res}\n```\n"
 
 def is_litellm_model(model):
+    from litellm import models_by_provider
     provider = None
     model_name = None
     if model.startswith ("/"):
@@ -98,7 +97,7 @@ def is_litellm_model(model):
         provider, model_name = model.split(":")
     elif "/" in model:
         provider, model_name = model.split("/")
-    if provider in litellm.models_by_provider and model_name in litellm.models_by_provider[provider]:
+    if provider in models_by_provider and model_name in models_by_provider[provider]:
         return True
     return False
 
@@ -378,6 +377,7 @@ def respond(self):
         # builtins.print(prompt)
         response = None
         if self.auto_run:
+            from . import auto
             if(is_litellm_model(self.model)):
                 response = auto.chat(self)
             else:
@@ -416,7 +416,8 @@ def respond(self):
             #  {"role": "system", "content": "You are a poetic assistant, be creative."},
             #  {"role": "user", "content": "Compose a poem that explains the concept of recursion in programming."}
             # ]
-            completion = litellm.completion(
+            from litellm import completion as litellm_completion
+            completion = litellm_completion(
                 model=self.model.replace(":", "/"),
                 messages=self.messages,
                 max_completion_tokens=maxtokens,
@@ -452,6 +453,7 @@ def respond(self):
                     "max_tokens": maxtokens
                 }
                 if self.env["chat.rawdog"] == "true":
+                    from .completion import messages_to_prompt
                     prompt = messages_to_prompt(self, messages)
                     response = self.llama_instance(prompt, **chat_args)
                 else:

r2ai/models.py

@@ -1,20 +1,14 @@
 from .utils import slurp, dump
-from huggingface_hub import hf_hub_download, login
-from huggingface_hub import HfApi, list_repo_tree, get_paths_info
 from typing import Dict, List, Union
 import appdirs
 import builtins
 import inquirer
 import json
-import llama_cpp
 import os
 import shutil
 import subprocess
 import sys
 import traceback
-from transformers import AutoTokenizer
-from llama_cpp.llama_tokenizer import LlamaHFTokenizer
-
 # DEFAULT_MODEL = "TheBloke/CodeLlama-34B-Instruct-GGUF"
 # DEFAULT_MODEL = "TheBloke/llama2-7b-chat-codeCherryPop-qLoRA-GGUF"
 # DEFAULT_MODEL = "-m TheBloke/dolphin-2_6-phi-2-GGUF"
@@ -263,6 +257,7 @@ def get_hf_llm(ai, repo_id, debug_mode, context_window):
 
             # Check if model was originally split
             split_files = [model["filename"] for model in raw_models if selected_model in model["filename"]]
+            from huggingface_hub import hf_hub_download
             if len(split_files) > 1:
                 # Download splits
                 for split_file in split_files:
@@ -466,6 +461,7 @@ def list_gguf_files(repo_id: str) -> List[Dict[str, Union[str, float]]]:
     """
 
     try:
+      from huggingface_hub import HfApi
       api = HfApi()
       tree = list(api.list_repo_tree(repo_id))
       files_info = [file for file in tree if file.path.endswith('.gguf')]
@@ -634,7 +630,9 @@ def supports_metal():
 
 
 def get_llama_inst(repo_id, **kwargs):
+    import llama_cpp
     if 'functionary' in repo_id:
+        from llama_cpp.llama_tokenizer import LlamaHFTokenizer
         kwargs['tokenizer'] = LlamaHFTokenizer.from_pretrained(repo_id)
         filename = os.path.basename(kwargs.pop('model_path'))
         kwargs['echo'] = True

r2ai/pipe.py

@@ -1,7 +1,7 @@
 import os
 import traceback
 import r2pipe
-from .progress import progress_bar
+
 
 have_rlang = False
 r2lang = None
@@ -63,7 +63,6 @@ def r2singleton():
 def get_r2_inst():
     return r2singleton()
 
-@progress_bar("Loading", color="yellow")
 def open_r2(file, flags=[]):
     global r2, filename, r2lang
     r2 = r2pipe.open(file, flags=flags)