transformers
transformers copied to clipboard
huggingface
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.
# What does this PR do? Fixes https://github.com/huggingface/transformers/issues/36110 This PR adds [DeepseekAI](https://deepseek.ai/)'s DeepseekVL model to Hugging Face Transformers. DeepseekVL is an open-source Vision-Language (VL) Model designed for real-world vision and...
# What does this PR do? Fixes #26286 Solves the duplicate answers problem when `align_to_words=True` which is default in the `question_answering` pipeline by keeping track of the answers history and...
### System Info Environment Details: • transformers version: 4.51.3 • tokenizers version: (e.g., 0.21.1 or the version pip installed with transformers 4.51.3) • torch version: 2.7.0 (with MPS available and...
### System Info Master branch The result of `speed_metrics` is not logged or saved https://github.com/huggingface/transformers/blob/4005e30c804f9b3a9dbf45d019f6ca1cdc4d774c/src/transformers/trainer.py#L3656 ### Who can help? @zach-huggingface @SunMarc ### Information - [ ] The official example scripts...
> @SSacSim Yes One have to import cv2 and pop up window frist. Thats really annoying. > >  > > > transformers 4.48.3 > opencv-python 4.10.0.84 > > @purusharthmalik...
# What does this PR do? Fixes # (issue) ## Before submitting - [ ] This PR fixes a typo or improves the docs (you can dismiss the other checks...
Add `SepCache` [An efficient and easy-to-use Cache from the SepLLM paper - ICML 2025 (https://arxiv.org/abs/2412.12094) ] to the `cache_utils.py` and `__init__.py` A cache as described in the [SepLLM paper -...
in Trainer.prediction_step(...) ```python def prediction_step( self, model: nn.Module, inputs: dict[str, Union[torch.Tensor, Any]], prediction_loss_only: bool, ignore_keys: Optional[list[str]] = None, ) -> tuple[Optional[torch.Tensor], Optional[torch.Tensor], Optional[torch.Tensor]]: has_labels = False if len(self.label_names) == 0...
# What does this PR do? Video inference broke after https://github.com/huggingface/transformers/pull/38105 because the model was inferring number of frames before processing. Previously that worked fine, but with the new workflow...
Add FAST
# What does this PR do? This PR adds FAST: Faster Arbitrarily-Shaped Text Detector with Minimalist Kernel Representation. It should be merged after the first pr for its backbone, textnet,...
Metadata
Owner
Metadata
🤗 Transformers: State-of-the-art Machine Learning for Pytorch, TensorFlow, and JAX.