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
Thank you for your amazing work.I searched online and found that the CLIP-ViT-L/14@336px model divides an image into 14*14=196 patches, and the embedding dimension is 768. In your work the shape of features after CLIP visual encoder is (576,1024). How does it come?
The text was updated successfully, but these errors were encountered:
Thank you for your amazing work.I searched online and found that the CLIP-ViT-L/14@336px model divides an image into 14*14=196 patches, and the embedding dimension is 768. In your work the shape of features after CLIP visual encoder is (576,1024). How does it come?
The text was updated successfully, but these errors were encountered: