Autore | SHA1 Messaggio | Data |
---|---|---|
|
e26a4ac698 chore: avoid loading the unused layers and init the VLM up to the required feature space | 5 mesi fa |
|
c11a8bdaad fix: calculate max number of multi-modal tokens automatically | 5 mesi fa |
|
4599c98f99 feat: dynamic image size support for VLMs | 5 mesi fa |
|
3a0fdf7b9b chore: remove `image_input_type` from VLM config | 6 mesi fa |
|
c0c336aaa3 refactor: registry for processing model inputs; quick_gelu; clip model support | 6 mesi fa |