transformers
4c040aba - [mistral] Support passing `head_dim` through config (and do not require `head_dim * num_heads == hidden_size`) (#32050)

Commit

1 year ago

[mistral] Support passing `head_dim` through config (and do not require `head_dim * num_heads == hidden_size`) (#32050) * Allow `head_dim` to be set in Mistral config * Add docstring * Do not require `head_dim * num_heads == hidden_size` * [run-slow] mistral

References

#29969 - [SigLIP] Add fast tokenizer

#32050 - [mistral] Support passing `head_dim` through config (and do not require `head_dim * num_heads == hidden_size`)

#32831 - [Docs] Update resources

#33111 - [Backbone] Remove out_features everywhere

#33174 - [Zero-shot image classification pipeline] Remove tokenizer_kwargs

#39821 - Support MetaCLIP 2

#58 - Add EoMT DINOv3 model

#59 - Fix attention mask handling in EoMT-DINOv3 converter

#41212 - Add EoMT with DINOv3 backbone

#62 - Add initial DEIMv2 model implementation

Author

xenova

Parents

c50e0551

transformers 4c040aba - [mistral] Support passing `head_dim` through config (and do not require `head_dim * num_heads == hidden_size`) (#32050)

transformers
4c040aba - [mistral] Support passing `head_dim` through config (and do not require `head_dim * num_heads == hidden_size`) (#32050)