Sigmoid Loss for Language Image Pre-Training
Paper
•
2303.15343
•
Published
•
5
SigLIP release
SigLIP improves upon CLIP with a sigmoid loss. Both English-only and multilingual checkpoints are released.
-
-
google/siglip-base-patch16-224
Zero-Shot Image Classification • Updated • 766k • 30 -
google/siglip-base-patch16-256
Zero-Shot Image Classification • Updated • 10.8k • 5 -
google/siglip-base-patch16-384
Zero-Shot Image Classification • Updated • 245k • 9 -
google/siglip-base-patch16-512
Zero-Shot Image Classification • Updated • 61.6k • 17 -
google/siglip-large-patch16-256
Zero-Shot Image Classification • Updated • 52.7k • 10 -
google/siglip-large-patch16-384
Zero-Shot Image Classification • Updated • 191k • 6
google/siglip-base-patch16-256-multilingual
Zero-Shot Image Classification • Updated • 48.2k • 37Note This is a multilingual variant trained on WebLi without language filtering.
google/siglip-so400m-patch14-384
Zero-Shot Image Classification • Updated • 2.92M • 386Note This one has a scaled-optimized architecture, obtaining the best performance in English.