You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
I attempted to generate CAMs, but due to the different model structure, I am unsure if I have implemented it as you intended. (The results were not as good as expected, so I am reaching out for help.)
Could you provide a clip_wrapper.py that is compatible with open_clip? (I tried it myself, but I am keen to understand your precise implementation.)
First of all, thank you for your excellent research.
I am interested in applying your approach using a CLIP model trained in a different domain.
I would like to know how to obtain CAM results from a pretrained CLIP using open-clip (https://github.com/mlfoundations/open_clip) instead of OpenAI's CLIP.
I attempted to generate CAMs, but due to the different model structure, I am unsure if I have implemented it as you intended. (The results were not as good as expected, so I am reaching out for help.)
Could you provide a clip_wrapper.py that is compatible with open_clip? (I tried it myself, but I am keen to understand your precise implementation.)
I have attached the network in a txt file.
open-clip.txt
open_ai_clip.txt
The text was updated successfully, but these errors were encountered: