Pablo Montalvo

Raise unused kwargs image processor

3

# What does this PR do? This PR captures all kwargs passed to an `ImageProcessor` `preprocess` method and compares them to what's expected, raising an exception or logging an informative...

[WIP] Improve multimodal processors - rely less on kwargs

1

# What does this PR do? This PR aims at a better control on the logic flow through `Processor` classes, in particular those leveraging `ImageProcessor` with a `Tokenizer`. Linked with...

Abstract image processor arg checks.

3

# What does this PR do? This refactors existing image processor argument checks that sprawl out on all existing models that have an `ImageProcessor`. Lines such as ```python if do_resize...

Image + text + audio uniform processors

3

# What does this PR do? This PR is a stab at uniformizing the processors across all transformers models. If we are happy with the design, I'll expand it to...

Paligemma- fix devices and dtype assignments

2

# What does this PR do? Moves tensors to correct devices in case of multi-gpu training on accelerate and device_map = auto. Additionally ensures bf16 training works as well. Fixes...

Mamba 2 Multi-GPU errors out on generation with parallel beam search

1

### Observed issue Found out when running multi-gpu slow tests in https://github.com/huggingface/transformers/pull/33560 . Line 479 exactly of the mamba2 modeling file https://github.com/huggingface/transformers/blob/8efc06ee1863bd6e34e8adb7b10901da87c66818/src/transformers/models/mamba2/modeling_mamba2.py#L472-L480 Will raise the following for the test `tests/models/mamba2/test_modeling_mamba2.py::Mamba2ModelTest::test_model_parallel_beam_search`...