transformers
d4bd33cc - Register ModelOutput subclasses as supported torch.utils._pytree nodes (#25358)

Commit

2 years ago

Register ModelOutput subclasses as supported torch.utils._pytree nodes (#25358) * Register ModelOutput subclasses as supported torch.utils._pytree nodes Fixes #25357 where DDP with static_graph=True does not sync gradients when calling backward() over tensors contained in ModelOutput subclasses * Add test for torch pytree ModelOutput serialization and deserialization

References

#25358 - Register ModelOutput subclasses as supported torch.utils._pytree nodes

#27720 - Add common processor tests

#29969 - [SigLIP] Add fast tokenizer

#32831 - [Docs] Update resources

#33111 - [Backbone] Remove out_features everywhere

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

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

#62 - Add initial DEIMv2 model implementation

#65 - Fix RTDetrV2 sine position embedding ordering

#44375 - Add RF-DETR

#71 - Use Mask2Former ignore_value in mask matching and losses