riaz.somc/transformers
1
0
Fork 0
mirror of https://github.com/huggingface/transformers.git synced 2025-07-18 03:58:25 +06:00
Code Issues Actions 23 Packages Projects Releases Wiki Activity
19,383 Commits 876 Branches 227 Tags 2 GiB
remove-script-datasets-in-tests-test-datasets-main
Commit Graph

19383 Commits

This Branch
This Branch
All Branches
Author SHA1 Message Date
Cyril Vallez
ad30598923
Update Mistral converter (#35967) 
* Update convert_mistral_weights_to_hf.py

* Update convert_mistral_weights_to_hf.py

* update

* style

* move it to integrations

* style

* trigger CIs

* trigger CIs
 2025-02-04 11:13:12 +01:00
Ryoo Kwangrok
b1954fd64a
layernorm_decay_fix (#35927) 
* layernorm_decay_fix

* W293 fix

* ruff format fix

* black format

* ruff format

* erase last layer

* add test_get_parameter_names_rmsnorm

* rmsnorm fix
 2025-02-04 11:01:49 +01:00
Dmitry Tarasov
2ba040a71f
apply_chat_template: consistent behaviour for return_assistant_tokens_mask=True return_tensors=True (#35582) 
* apply_chat_template: consistent return_tensors behaviour with return_assistant_tokens_mask flag

* test_chat_template_return_assistant_tokens_mask: support tokenizers with no attention mask

* test_chat_template_return_assistant_tokens_mask: skip tokenizers with no padding token

* test_chat_template_return_assistant_tokens_mask: force tokenizer padding_side=right

---------

Co-authored-by: Eduard Allakhverdov <goncharova@airi.net>
Co-authored-by: d.tarasov <d.tarasov@airi.net>
 2025-02-04 10:27:52 +01:00
Pavel Iakubovskii
9c02cb6233
Fix custom kernel for DeformableDetr, RT-Detr, GroindingDINO, OmDet-Turbo in Pytorch 2.6.0 (#35979) 
Updates type().is_cuda() -> .is_cuda(); .data<> -> .data_ptr<>
 2025-02-04 09:07:25 +00:00
Raushan Turganbay
5d75a25b03
Qwen2-VL: fix rope delta calculation (#36013) 
* fix rope delats calculation

* add test

* style
 2025-02-04 09:48:29 +01:00
Alex Brooks
e284c7e954
Update Granite Vision Model Path / Tests (#35998) 
* Update granite vision model path

Signed-off-by: Alex-Brooks <Alex.brooks@ibm.com>

* Enable granite vision test

Signed-off-by: Alex-Brooks <Alex.brooks@ibm.com>

---------

Signed-off-by: Alex-Brooks <Alex.brooks@ibm.com>
 2025-02-03 20:06:03 +01:00
Gar
9d2056f12b
Add mean_resizing for every VLMs' resizing_token_embeddings() (#35717) 
* refine all resize_token_embedding()

* ruff format

* hotfix
 2025-02-03 15:03:49 +01:00
Arthur
7eecdf2a86
Update-tp test (#35844) 
* update test for now

* up

* cleanup

* update todo
 2025-02-03 09:37:02 +01:00
Yih-Dar
62db3e6ed6
use torch 2.6 for daily CI (#35985) 
use torch 2.6 for CI

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
 2025-01-31 18:58:23 +01:00
Yoni Gozlan
2b46943195
Add GOT-OCR 2.0 to Transformers (#34721) 
* init modular got_ocr2

* Get correct got_ocr architecture

* add processing

* run modular with processing

* add working inference

* apply modular

* Refactor and fix style

* Refactor, cleanup, fix style

* fix init order

* Fix docs

* add base modeling tests

* fix style and consistency

* rename doc file

* fix repo consistency

* fix inference with box

* add image processing and support for crop_to_multi_page

* Fix batch inference

* add tests

* fixup

* fix slow test

* fix docstrings

* Add model doc

* update to new init

* fix input autocast pixel_values dtype

* update doc

* move doc to multimodal

* Reformat crop_image_to_patches and add docstrings

* Fix example in forward docstring

* Address Pablo review

* [run slow] got_ocr2

* remove defaults defined twice

* apply modular

* add torch_device to integration tests

* update modular

* follow-up Pavel review

* add device variable in doc

* fix doc multi-page

* Force eager attention for vision encoder to avoid attn implementation conflict

* revert qwen2vl doc changes

* use Qwen2ForCausalLM instead of Qwen2Model

* make fixup

* refactor gotocr2 to llava style

* uniformize function names and reduce checks

* final nits

* fix pixel_values dtype error

* change checkpoint names

* fix modular
 2025-01-31 11:28:13 -05:00
Joao Gante
5bbee12ac9
[Moshi] disable automatic compilation if the model can't compile (#35992) 
moshi cant compile
 2025-01-31 15:53:06 +00:00
eustlb
e6f4a4ebbf
[Moonshine] compute head_dim_padding at init (#35984) 
compute head_dim_padding at init
 2025-01-31 14:26:52 +01:00
Yoni Gozlan
d7188ba600
Add support for nested images to LLava and VipLLava (#35558) 
* move make_flat_list_of_images and make_batched_videos to image_utils

* remove unnecessary is_vision_available

* move make_nested_list_of_images to image_utils

* fix fast pixtral image processor

* fix import mllama

* fix make_nested_list_of_images

* add tests

* convert 4d arrays/tensors to list

* add test_make_batched_videos

* add support nested batch of videos

* fix image processing qwen2vl
 2025-01-30 16:49:20 -05:00
Marcel
e4227eb4d4
Handle empty change indices in SAM's mask to rle conversion (#35665) 
* Handle empty change indices in RLE conversion for masks

* [test] Add unit tests for RLE encoding of masks in SamProcessor

* [test] Update RLE conversion tests to use TensorFlow implementation

* [test] Fix formatting in SamProcessorTest according to check_code_quality action

* [test] Fix formatting in SamProcessorTest according to check_code_quality

* [test] Refactored rle test cases into one test and used tf tensors in tf test cases

* [test] Fix: removed self parameter from refactored methods

* [test] Removed nested methods in run-length encoding tests for PyTorch and TensorFlow

* [test] Added description to individual to run-length encoding tests for PyTorch and TensorFlow.
 2025-01-30 19:08:38 +00:00
Yih-Dar
47bd4296d6
not to use A100 for benchmark.yml (#35974) 
fix

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
 2025-01-30 18:55:36 +01:00
Nat Jeffries
693328f2bc
Support batching for UsefulSensors Moonshine (#35922) 
* Add support for attention masking in moonshine.

Tested against Open ASR Leaderboard with batch size 256.

* Update comments and ensure attention masks are passed everywhere.

Perform attention mask downsampling inside of moonshine forward call.

* Hide padding behind conditional. Fix encoder/decoder masking.

- Correctly pipe encoder attention mask into decoder
- Add correct scaling factor if one is not already provided.
- Fix formatting with ruff

* Add auto generated modeling_moonshine file.

* Update formatting in generated model file.

* Address review comments.

* Fix typo.

* Add `pad_head_dim_to_multiple_of` to moonshine config.

* Correct args order for MooonshineConfig.

* Update configuration moonshine too.

* Update src/transformers/models/moonshine/modular_moonshine.py

* Update src/transformers/models/moonshine/configuration_moonshine.py

---------

Co-authored-by: eustlb <94853470+eustlb@users.noreply.github.com>
 2025-01-30 17:08:07 +01:00
Yih-Dar
5757681837
Less flaky for TimmBackboneModelTest::test_batching_equivalence (#35971) 
* fix

* remove is_flaky

* fix

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
 2025-01-30 16:56:26 +01:00
Matt
e320d5542e
Revert p_mask to a list in DQA pipeline (#35964) 
* p_mask back to being a list

* Remove breakpoint
 2025-01-30 15:37:59 +00:00
Raushan Turganbay
365fecb4d0
Whisper: fix static cache CI (#35852) 
* fix

* remove overriden method

* small change
 2025-01-30 12:43:00 +01:00
Raushan Turganbay
9725e5be2f
Pixtral: vectorize patch embeddings and enable tests (#35122) 
* initial POC

* - batch mix feature

* fix tests

* fix tests

* make style

* do not skip and instead fix tests

* update

* return back the test

* correct text with the correct ckpt
 2025-01-30 12:40:18 +01:00
Joao Gante
8bc4c89ee9
[bart] minor test fixes (#35965) 
fix tests
 2025-01-30 10:00:11 +00:00
Ilyas Moutawwakil
19f2ec80cf
Fix is_causal being a tensor (#35791) 
* fix is_causal being a tensor

* convert in sdpa attention only when  jit tracing
 2025-01-30 09:22:33 +01:00
Wing Lian
7547f55e5d
fix iterator overflow when gradient accumulation is 1 (#35960) 2025-01-29 14:45:09 -05:00
Joao Gante
4d3b1076a1
[generate] move max time tests (#35962) 
* move max time tests to their right place

* move test to the right place
 2025-01-29 17:56:46 +00:00
Boris Malashenko
4d1d489617
Update README.md (#35958) 
There should be a dot after pip install .
 2025-01-29 15:46:26 +00:00
Fanli Lin
f0ae65c198
[tests] further fix Tester object has no attribute '_testMethodName' (#35781) 
* bug fix

* update with more cases

* more entries

* Fix

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
 2025-01-29 16:05:33 +01:00
Yih-Dar
ec7790f0d3
update docker file transformers-pytorch-deepspeed-latest-gpu (#35940) 
update docker file for deepspeed

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
 2025-01-29 16:01:27 +01:00
Zach Mueller
5d257111c1
Trainer Refactor: Part 1 (#35567) 
* start

* So far: 30%

* Small fix

* Continuing update

* Continuing

* Forgot to check if not None

* Continuing refactor

* Fix if else

* Fix ref

* Should make tests pass

* Keep grad norm same

* Document

* Apply suggestions from code review

Co-authored-by: Marc Sun <57196510+SunMarc@users.noreply.github.com>

* Err instead of info for logging RNG state error

* Seperate out to func

---------

Co-authored-by: Marc Sun <57196510+SunMarc@users.noreply.github.com>
 2025-01-29 09:50:54 -05:00
Jonas Rohw
23d782ead2
Output dicts support in text generation pipeline (#35092) 
* Support for generate_argument: return_dict_in_generate=True, instead of returning a error

* fix: call test with return_dict_in_generate=True

* fix: Only import torch if it is present

* update: Encapsulate output_dict changes

* fix: added back original comments

---------

Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>
 2025-01-29 14:44:46 +00:00
Yih-Dar
cf90404807
Fix flaky test_assisted_decoding_matches_greedy_search (#35951) 
fix

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
 2025-01-29 14:50:07 +01:00
Yih-Dar
692afa102d
Update squad_convert_example_to_features to work with numpy v2 (#35955) 
* Fix

* Fix

* Fix

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
 2025-01-29 14:33:06 +01:00
Yih-Dar
c600e89f5c
Update unwrap_and_save_reload_schedule to use weights_only=False (#35952) 
* fix

* Fix

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
 2025-01-29 14:30:57 +01:00
Nadav Timor
42c8ccfd4c
fix test_generated_length_assisted_generation (#34935) 
fix test_generated_length_assisted_generation
 2025-01-29 12:03:45 +00:00
Mohamed Abu El-Nasr
ec7afad609
use torch constraints to check if covariance is positive definite during mean resizing. (#35693) 
* use torch constraints to check for psd

* small nit

* Small change

* Small change for the ci

* nit
 2025-01-28 17:33:42 +01:00
Ella Charlaix
61cbb723fc
Remove INC notebook reference in documentation (#35936) 
remove INC notebook in documentation
 2025-01-28 17:10:02 +01:00
NanoCode012
478c4f2d0d
fix(FA): QKV not being casted to target_dtype for FA with dpo lora (#35834) 
fix(FA): QKV not being casted to target_dtype due to dtype check
 2025-01-28 17:06:56 +01:00
Joao Gante
ece8c42488
Test: generate with torch.compile(model.forward) as a fast test (#34544) 2025-01-28 14:10:38 +00:00
Cyril Vallez
f48ecd7608
Fix TP initialization (#35860) 
* fix tp

* Update modeling_utils.py

* style

* style

* Update test_tp.py

* Update test_tp.py

* style

* Update test_tp.py

* Update test_tp.py

* Update test_tp.py

* Update test_tp.py
 2025-01-28 15:07:37 +01:00
Raushan Turganbay
f85ba20449
Qwen-2-5-VL: fix CI (#35935) 
fix
 2025-01-28 14:51:57 +01:00
Cyril Vallez
3f860dba55
Fix mask slicing for models with HybridCache (#35681) 
* correctly slice

* check mask

* Update modular_gemma2.py

* fix

* add tests

* fix typo

* finally fix mask slicing

* Finally correctly slice in all cases!!

* add test for all attention functions

* small fix in tests

* trick around dynamo tracing issue

* last update

* more robust

* kwargs propagation

* make it explicit for checkpointing

* apply modular
 2025-01-28 14:35:00 +01:00
Raushan Turganbay
b764c20b09
Fix: loading DBRX back from saved path (#35728) 
* fix dtype as dict for some models + add test

* add comment in tests
 2025-01-28 11:38:45 +01:00
Cyril Vallez
3613f568cd
Add default TP plan for all models with backend support (#35870) 
* Add some tp plans!

* More tp plans!

* Add it in the comment

* style

* Update configuration_mixtral.py

* Update configuration_phi.py

* update the layout according to special archs

* fix mixtral

* style

* trigger CIs

* trigger CIs

* CIs

* olmo2

---------

Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
 2025-01-28 11:20:58 +01:00
ivarflakstad
96625d85fd
Use rocm6.2 for AMD images (#35930) 
* Use rocm6.2 as rocm6.3 only has nightly pytorch wheels atm

* Use stable wheel index for torch libs
 2025-01-28 11:10:28 +01:00
Yih-Dar
bf16a182ba
Remove _supports_static_cache = True for some model classes (#34975) 
* use mask_fill

* remove comment

---------

Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
 2025-01-28 10:42:10 +01:00
Steven Liu
86d7564611
[docs] Fix Zamba2 (#35916) 
fix code block
 2025-01-27 11:44:10 -08:00
Matt
414658f94f
Close Zamba2Config code block (#35914) 
* close zamba2 code block

* Add Zamba2 to toctree
 2025-01-27 19:09:42 +00:00
Matt
63e9c941eb
Fix the config class comparison for remote code models (#35592) 
* Fix the config class comparison when repeatedly saving and loading remote code models

* once again you have committed your debug breakpoint
 2025-01-27 18:37:30 +00:00
Steven Liu
c550a1c640
[docs] uv install (#35821) 
uv install
 2025-01-27 08:49:28 -08:00
CalOmnie
cd6591bfb2
Fix typing in audio_utils.chroma_filter_bank (#35888) 
* Fix typing in audio_utils.chroma_filter_bank

* Apply make style

---------

Co-authored-by: Louis Groux <louis.cal.groux@gmail.com>
 2025-01-27 16:06:03 +00:00
Isotr0py
e57b459997
Split and clean up GGUF quantization tests (#35502) 
* clean up ggml test

Signed-off-by: Isotr0py <2037008807@qq.com>

* port remaining tests

Signed-off-by: Isotr0py <2037008807@qq.com>

* further cleanup

Signed-off-by: Isotr0py <2037008807@qq.com>

* format

Signed-off-by: Isotr0py <2037008807@qq.com>

* fix broken tests

Signed-off-by: Isotr0py <2037008807@qq.com>

* update comment

Signed-off-by: Isotr0py <2037008807@qq.com>

* fix

Signed-off-by: Isotr0py <2037008807@qq.com>

* reorganize tests

Signed-off-by: Isotr0py <2037008807@qq.com>

* k-quants use qwen2.5-0.5B

Signed-off-by: Isotr0py <2037008807@qq.com>

* move ggml tokenization test

Signed-off-by: Isotr0py <2037008807@qq.com>

* remove dead code

Signed-off-by: Isotr0py <2037008807@qq.com>

* add assert for serilization test

Signed-off-by: Isotr0py <2037008807@qq.com>

* use str for parameterize

Signed-off-by: Isotr0py <2037008807@qq.com>

---------

Signed-off-by: Isotr0py <2037008807@qq.com>
 2025-01-27 15:46:57 +01:00