Use clip to load local model

[tian-003]

There is an error when I use clip to load the local model, my code is as follows:
.map('img', 'vec', ops.image_text_embedding.clip(model_name='clip_vit_base_patch16', modality='image', checkpoint_path='./model/clip-vit-base-patch16/pytorch_model.bin').

The error that occurs is as follows:
File "/root/.towhee/operators/image-text-embedding/clip/versions/main/clip.py", line 108, in init
self.model = Model(real_name, modality, checkpoint_path, device)
File "/root/anaconda3/envs/towhee_env/lib/python3.7/site-packages/towhee/runtime/runtime_conf.py", line 88, in _decorated
return model(*args, **kwargs)
File "/root/.towhee/operators/image-text-embedding/clip/versions/main/clip.py", line 79, in init
self.model = create_model(model_name, modality, checkpoint_path, device)
File "/root/.towhee/operators/image-text-embedding/clip/versions/main/clip.py", line 44, in create_model
hf_clip_config = CLIPModel. from_config(model_name)

AttributeError: type object 'CLIPModel' has no attribute 'from_config'

How to solve this problem? How do i load local models?

[0xac9527]

change load clipmodel codes in clip.py (/root/.towhee/operators/image-text-embedding/clip/versions/main/clip.py)
e.g. hf_clip_config = CLIPModel.from_config(model_name) - > hf_clip_model = CLIPModel.from_pretrained(checkpoint_path)

that works