* just squash into one commit

* delete print

* enable csm test cases on XPU, all passed

Signed-off-by: Matrix Yao <matrix.yao@intel.com>

* fix style

Signed-off-by: Matrix Yao <matrix.yao@intel.com>

---------

Signed-off-by: Matrix Yao <matrix.yao@intel.com>

fix tp plan

* Fix incorrect attention mask truncate in whisper flash attention

* also fix incorrect attention mask truncate in qwen2 audio

* Nit attention mask truncate modeling_qwen2_audio.py

* Nit attention mask truncate modeling_whisper.py

Co-authored-by: Anton Vlasjuk <73884904+vasqu@users.noreply.github.com>

---------

Co-authored-by: Anton Vlasjuk <73884904+vasqu@users.noreply.github.com>
Co-authored-by: eustlb <94853470+eustlb@users.noreply.github.com>

add pre_output in the front

Co-authored-by: Pavel Iakubovskii <qubvel@gmail.com>

Add manueldeprada to run_slow allowed users

Fix typo in torch.compile method parameters

* Add config validation and style tweaks

* Fix style issues

* Fix style issues

* style

* Small fixes for copy/paste errors

---------

Co-authored-by: Cyrile <cyrile.delestre@arkea.com>

Ensure --auto_find_batch_size is the last test arg so indexing is correct

Fix temporal padding in Qwen2VLImageProcessor when the number of frames is not divisible by temporal_patch_size (#38076)

Qwen2VL: Fix temporal padding in Qwen2VLImageProcessor when frames are not divisible by temporal_patch_size

* fix tests

* delete

* fix one more test

* fix qwen + some tests are failing irrespective of `VideoProcessor`

* delete file

* enable finegrained_fp8 cases on XPU

Signed-off-by: Yao Matrix <matrix.yao@intel.com>

* fix style

Signed-off-by: Yao Matrix <matrix.yao@intel.com>

* change back to auto

Signed-off-by: Yao Matrix <matrix.yao@intel.com>

* rename per comments

Signed-off-by: Matrix Yao <matrix.yao@intel.com>

---------

Signed-off-by: Yao Matrix <matrix.yao@intel.com>
Signed-off-by: Matrix Yao <matrix.yao@intel.com>
Co-authored-by: Marc Sun <57196510+SunMarc@users.noreply.github.com>

* Update stopping_criteria.py

Fix description and formatting errors.

* Update stopping_criteria.py

Align formatting with existing files for consistency.

add style bot

update test for t4 runners

* init vilt image processor fast

* Refactor image processor tests to use loop for all processors

* Add ViltImageProcessorFast with PyTorch-based optimized image processing

* Change made automatically by make fixup command

* Change made automatically by make fix-copies command

* Fix type hints in ViltImageProcessorFast for Python compatibility

* Define constants for image resizing based on COCO dataset aspect ratio

* Add missing property initializations to ViltImageProcessorFast

* Extract resize logic into dedicated method in ViltImageProcessorFast

* Extract padding logic into dedicated method

* Implement shape-based image grouping for optimized processing in Vilt

* Update test suite to verify ViltImageProcessorFast attributes

* Move variable declarations to _preprocess method parameters

* Remove unused parameters

* Rename _resize method to resize to override existing function

* Remove whitespace

* Remove unnecessary type check and conversion for stacked_images

* Remove redundant loop and apply padding directly to stacked images

* Refactor pad function to return images and mask as tuple instead of dict

* Add tests comparing padding masks in slow and fast implementations

* Update ViltImageProcessor tests to ensure compatibility between slow and fast implementations

* Replace add_start_docstrings with auto_docstring in ViltImageProcessorFast

* Move docstrings of custom args to ViltFastImageProcessorKwargs

* Use reorder_images function for both masks and images

---------

Co-authored-by: Yoni Gozlan <74535834+yonigozlan@users.noreply.github.com>

* fix InternVL interpolate_pos_encoding

* fix modular and auto_video_processor for internvl

fix

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

* fix llava processor to calculate unpad size correctly

* repo consistency

* Revert "repo consistency" & "setUp in llava family"

This reverts commit 26a50af8db5b15bb6b700db3d53342fe69579d8e.

* add edge case test for padding & unpadding

* compute unpadding size from original size

* make test config explicit

* Revert "compute unpadding size from original size"

This reverts commit 752cd27ad9710ab056c17a9986760c4651975540.

* Revert "add edge case test for padding & unpadding"

This reverts commit ccbd094d69c3f8f6a259159164284f60ba835bce.

* revert unpad logic

* remove irrelevant tests

* model test

* remove processor from model test

---------

Co-authored-by: jaycha <jaycha@ncsoft.com>

* F: Fix type hint.

* F: Use Cache type.

* F: Sort import.

* U: Format.

* U: Address reviews.

Fix bug in prefill_chunk_size implementation that ignores disable_compile flag

skip the test

* Disable report callbacks for certain training tests

* Disable report callbacks for test_auto_batch_size_finder

fix: Propagate `lr_scheduler_kwargs` options to create LR Scheduler when LayerWiseDummyOptimizer is used (#34559)

fix: fix get_scheduler

* add timeout

* change 10 to 60

* update require_read_token

* new repo

* fix

* fix

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

* refactor image processor phi4

* nits fast image proc

* add image tests phi4

* Fix image processing tests

* update integration tests

* remove revision and add comment in integration tests

uninstall kernels

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

* update seed_worker to set seed based on worker_id and rank

* test case

* set output_dir as remove tmp dir

* fix total updates in epoch

* add test; fix max_steps

* replace with multi-gpu decorator

The commit (https://github.com/huggingface/transformers/commit/5cf11e5ab9591652ee025069658f9af5a98e455e) fixed the type hints for the parameter `tools` in apply_chat_template, but the docstring was not changed.

* chore(qwen2): display warning log only when sliding window attention is enabled

* Align modeling_qwen2.py and modular_qwen2.py

---------

Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* Fix integration tests

* format

* accept arbitrary kwargs

* move user commands to a separate fn

* work with generation config files

* rm cmmt

* docs

* base generate flag doc section

* nits

* nits

* nits

* no <br>

* better basic args description

* fix qwen2-vl loading

* fix a few nore models

* delete print

* fix copies