How to use rfigge/meru-base-open-ai-weights with OpenCLIP:
import open_clip model, preprocess_train, preprocess_val = open_clip.create_model_and_transforms('hf-hub:rfigge/meru-base-open-ai-weights') tokenizer = open_clip.get_tokenizer('hf-hub:rfigge/meru-base-open-ai-weights')
To use this, can I just load these weights into the CLIPModel from hugging face or do something like model = CLIPModel(configuration), where configuration is topen_clip_config.json? That way CLIP has the weights from the hyperbolic version: MERU?
Thank you!
· Sign up or log in to comment
