[Bug]: cannot import name '_encode_image' from 'ollama._client'

[zhangjy328]

Describe the bug

when i run train.py, ImportError: cannot import name '_encode_image' from 'ollama._client'

Dataset

Other (please specify in the text field below)

Model

PatchCore

Steps to reproduce the behavior

install anomalib use anaconda prompt

OS information

OS information:

• OS: win10
• Python version: 3.10
• Anomalib version: 1.2.0
• PyTorch version: 2.5.1+cpu
• CUDA/cuDNN version: none
• GPU models and configuration: cpu
• Any other relevant information: [e.g. I'm using a custom dataset]

Expected behavior

Traceback (most recent call last):
File "E:\model\anomalib-main\train.py", line 3, in
from anomalib.models import Patchcore
File "E:\model\anomalib-main\src\anomalib\models_init_.py", line 15, in
from .image import (
File "E:\model\anomalib-main\src\anomalib\models\image_init_.py", line 23, in
from .vlm_ad import VlmAd
File "E:\model\anomalib-main\src\anomalib\models\image\vlm_ad_init_.py", line 6, in
from .lightning_model import VlmAd
File "E:\model\anomalib-main\src\anomalib\models\image\vlm_ad\lightning_model.py", line 14, in
from .backends import Backend, ChatGPT, Huggingface, Ollama
File "E:\model\anomalib-main\src\anomalib\models\image\vlm_ad\backends_init_.py", line 9, in
from .ollama import Ollama
File "E:\model\anomalib-main\src\anomalib\models\image\vlm_ad\backends\ollama.py", line 23, in
from ollama._client import _encode_image
ImportError: cannot import name '_encode_image' from 'ollama._client' (C:\Users\zjy\AppData\Local\anaconda3\envs\anomalib_env\lib\site-packages\ollama_client.py)

Screenshots

No response

Pip/GitHub

GitHub

What version/branch did you use?

No response

Configuration YAML

# Import the required modules
from anomalib.data import MVTec
from anomalib.models import Patchcore
from anomalib.engine import Engine
from anomalib.data import Folder
from anomalib.utils.normalization import NormalizationMethod


# Initialize the datamodule, model and engine
# datamodule = MVTec(num_workers=0)
model = Patchcore()
engine = Engine(
    default_root_dir = './result',
    task = "classification",
    callbacks = None,
    normalization = NormalizationMethod.MIN_MAX,
    threshold = "F1AdaptiveThreshold",
    image_metrics = None,
    pixel_metrics = None,
    logger = None
)

datamodule = Folder(
    name="mvtec",
    root="../datasets/CSP",
    normal_dir="normal",
    abnormal_dir="abnormal",
    task="classification",
    train_batch_size=16,
    eval_batch_size=16,
    num_workers=0,
    image_size=[1024, 1024],
    mask_dir=None,
    normal_split_ratio=0.2,
    seed=None,
)

datamodule.setup()

# Train the model
engine.train(datamodule=datamodule, model=model)

Logs

Traceback (most recent call last):
  File "E:\model\anomalib-main\train.py", line 3, in <module>
    from anomalib.models import Patchcore
  File "E:\model\anomalib-main\src\anomalib\models\__init__.py", line 15, in <module>
    from .image import (
  File "E:\model\anomalib-main\src\anomalib\models\image\__init__.py", line 23, in <module>
    from .vlm_ad import VlmAd
  File "E:\model\anomalib-main\src\anomalib\models\image\vlm_ad\__init__.py", line 6, in <module>
    from .lightning_model import VlmAd
  File "E:\model\anomalib-main\src\anomalib\models\image\vlm_ad\lightning_model.py", line 14, in <module>
    from .backends import Backend, ChatGPT, Huggingface, Ollama
  File "E:\model\anomalib-main\src\anomalib\models\image\vlm_ad\backends\__init__.py", line 9, in <module>
    from .ollama import Ollama
  File "E:\model\anomalib-main\src\anomalib\models\image\vlm_ad\backends\ollama.py", line 23, in <module>
    from ollama._client import _encode_image
ImportError: cannot import name '_encode_image' from 'ollama._client' (C:\Users\zjy\AppData\Local\anaconda3\envs\anomalib_env\lib\site-packages\ollama\_client.py)

Code of Conduct

• I agree to follow this project's Code of Conduct

[samet-akcay]

Looks like the api has changed a bit in ollama. Can you try installing the library as follows: pip install "ollama<0.4"

[samet-akcay]

Until we adapt the new API changes, this should work

[zhangjy328]

but i have a new question, LightningModule.configure_optimizers returned None, this fit will run with no optimizer, and training_step returned None

C:\Users\zjy\AppData\Local\anaconda3\envs\anomalib_env\lib\site-packages\timm\models\layers_init_.py:48: FutureWarning: Importing from timm.models.layers is deprecated, please import via timm.layers
warnings.warn(f"Importing from {name} is deprecated, please import via timm.layers", FutureWarning)
INFO:anomalib.data.base.datamodule:No normal test images found. Sampling from training set using a split ratio of 0.20
dict_keys(['image_path', 'label', 'image'])
dict_keys(['image_path', 'label', 'image'])
dict_keys(['image_path', 'label', 'image'])
INFO:anomalib.models.components.base.anomaly_module:Initializing Patchcore model.
INFO:timm.models._builder:Loading pretrained weights from Hugging Face hub (timm/wide_resnet50_2.racm_in1k)
INFO:timm.models._hub:[timm/wide_resnet50_2.racm_in1k] Safe alternative available for 'pytorch_model.bin' (as 'model.safetensors'). Loading weights using safetensors.
INFO:timm.models._builder:Missing keys (fc.weight, fc.bias) discovered while loading pretrained weights. This is expected if model is being adapted.
GPU available: False, used: False
TPU available: False, using: 0 TPU cores
HPU available: False, using: 0 HPUs
INFO:anomalib.data.base.datamodule:No normal test images found. Sampling from training set using a split ratio of 0.20
WARNING:anomalib.metrics.f1_score:F1Score class exists for backwards compatibility. It will be removed in v1.1. Please use BinaryF1Score from torchmetrics instead
C:\Users\zjy\AppData\Local\anaconda3\envs\anomalib_env\lib\site-packages\lightning\pytorch\core\optimizer.py:182: LightningModule.configure_optimizers returned None, this fit will run with no optimizer

| Name | Type | Params | Mode

0 | model | PatchcoreModel | 24.9 M | train
1 | _transform | Compose | 0 | train
2 | normalization_metrics | MetricCollection | 0 | train
3 | image_threshold | F1AdaptiveThreshold | 0 | train
4 | pixel_threshold | F1AdaptiveThreshold | 0 | train
5 | image_metrics | AnomalibMetricCollection | 0 | train
6 | pixel_metrics | AnomalibMetricCollection | 0 | train

24.9 M Trainable params
0 Non-trainable params
24.9 M Total params
99.450 Total estimated model params size (MB)
15 Modules in train mode
174 Modules in eval mode
C:\Users\zjy\AppData\Local\anaconda3\envs\anomalib_env\lib\site-packages\lightning\pytorch\trainer\connectors\data_connector.py:424: The 'train_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of the num_workers argumenttonum_workers=11in theDataLoaderto improve performance. C:\Users\zjy\AppData\Local\anaconda3\envs\anomalib_env\lib\site-packages\lightning\pytorch\trainer\connectors\data_connector.py:424: The 'val_dataloader' does not have many workers which may be a bottleneck. Consider increasing the value of thenum_workers argument to num_workers=11 in the DataLoader to improve performance.
Epoch 0: 0%| | 1/250 [00:51<3:35:14, 0.02it/s]C:\Users\zjy\AppData\Local\anaconda3\envs\anomalib_env\lib\site-packages\lightning\pytorch\loops\optimization\automatic.py:132: training_step returned None. If this was on purpose, ignore this warning...

[samet-akcay]

Patchcore doesn't have an optimizer, so this is on purpose. As long as you get a val/test score, everything works as expected

[zhangjy328]

Thanks, So is everything normal like this? Just wait for it to finish training and produce results?

[zhangjy328]

And I also have another question：How can I continue training the model based on the results of the previous training session?

[neoragex2002]

Looks like the api has changed a bit in ollama. Can you try installing the library as follows: pip install "llama<0.4"

it is not pip install "llama<0.4"

it is pip install "ollama<0.4" ...

[samet-akcay]

well it's a typo, thanks for the fix