GitLab

prepare for "__floordiv__ is deprecated  and its behavior will change in a future version of pytorch" (#20211)

* rounding_mode = "floor"  instead of // to prevent behavioral change

* add other TODO

* use `torch_int_div` from pytrch_utils

* same for tests

* fix copies

* style

* use relative imports when needed

* Co-authored-by: sgugger <sylvain.gugger@gmail.com>

* Fix flaky test for log level

* Fix other flaky test

* First commit for the improved PT-TF weight loading

* Remove workarounds from TFEncoderDecoder tests

* Allow a custom weight renaming function in from_pretrained and use that to clean up EncoderDecoder

* make fixup

* First attempt at visionencoderdecoder

* Disable tensorfloat32 in tests to get consistent outputs

* Quick fix to tf_vision_encoder_decoder tests

* make fixup

* Update Blenderbot tests

* Remove unused arg in modeling_tf_opt

* load_tf_sharded_weights had strict=True! This meant transfer learning was impossible, so I'm setting it to False.

* Support prefixes when loading sharded TF checkpoints

* make fixup

* Add test to load sharded models with a weight prefix

* Fix sharded weight loading test

* Add a test for transfer from a sharded checkpoint

* make fixup

* Add test to check that crossloading from PT with a prefix works

* Refactor from_pretrained in the encoderdecoder classes

* Refactor from_pretrained in the encoderdecoder classes

* missmatched -> mismatched

* Explicitly check for None

* No comments showing my very impressive and attractive knowledge of Py3.9+

* Disable TF32 across all TF tests

* Add PipelineTesterMixin

* remove class PipelineTestCaseMeta

* move validate_test_components

* Add for ViT

* Add to SPECIAL_MODULE_TO_TEST_MAP

* style and quality

* Add feature-extraction

* update

* raise instead of skip

* add tiny_model_summary.json

* more explicit

* skip tasks not in mapping

* add availability check

* Add Copyright

* A way to diable irrelevant tests

* update with main

* remove disable_irrelevant_tests

* skip tests

* better skip message

* better skip message

* Add all pipeline task tests

* revert

* Import PipelineTesterMixin

* subclass test classes with PipelineTesterMixin

* Add pipieline_model_mapping

* Fix import after adding pipieline_model_mapping

* Fix style and quality after adding pipieline_model_mapping

* Fix one more import after adding pipieline_model_mapping

* Fix style and quality after adding pipieline_model_mapping

* Fix test issues

* Fix import requirements

* Fix mapping for MobileViTModelTest

* Update

* Better skip message

* pipieline_model_mapping could not be None

* Remove some PipelineTesterMixin

* Fix typo

* revert tests_fetcher.py

* update

* rename

* revert

* Remove PipelineTestCaseMeta from ZeroShotAudioClassificationPipelineTests

* style and quality

* test fetcher for all pipeline/model tests

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

* Add loss for BridgeTowerForMaskedLM and BridgeTowerForImageAndTextRetrieval

* minor fix return_dict

* implement test for loss computation

---------
Co-authored-by: Tiep Le <97980157+tileintel@users.noreply.github.com>
Co-authored-by: Tiep Le <tiep.le@intel.com>

* fix blip multi gpu

* fix

* final changes

* adapt suggestions

* fix failing slow test

* forward contrib credits from testing and suggestions

* reformat

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

* Use token

* Avoid failure

* better error

* Fix

* fix style

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

* zero shot object detection part 1

* added batch prediction section

* added image guided object detection section

* make style

* added the task guide to the TOC

* minor polishing

* Apply suggestions from code review
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
Co-authored-by: Alara Dirik <8944735+alaradirik@users.noreply.github.com>

* added embedded owlvit demo

* Apply suggestions from code review
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* minor fix

* make style

---------
Co-authored-by: Steven Liu <59462357+stevhliu@users.noreply.github.com>
Co-authored-by: Alara Dirik <8944735+alaradirik@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* If applied, this commit fixes generate bug in gptj

* Remove extra same code block

* formatting and test fix

* Conflict fix and declaration error fix

---------
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Fix the issue of blip model returning loss even when the label is not provoided

* Fix ruff failure

* Incorporate PR feedbacks

* Incorporate PR feedbacks

* Incorporate PR feedbacks

* Incorporate PR feedbacks

* add v1

* add `Blip2Model`

- add relevant functions
- add tests
- add on automapping

* fix docs

* fix doctest

* fix torchquant issue

* add tests

* fix tf random mask tokens probability

* fix tf random mask tokens probability in collator for langauge modelling

* Fix gradient checkpointing bug in gptneox

* Fix gradient checkpointing bug in modeling_imagegpt.py

* Revert gpt neox changes

---------
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

Co-authored-by: saswatmeher <saswatmeher@cse.iitb.ac.in>

Let's give TF a bit more love ❤️ 🙏

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

* logger.warning_once

* style

fix quality with ruff 0.0.253
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

* Fix gradient checkpointing bug in gptneox

* Remove use_cache block

fix nn.init.trunc_normal_ call on half data

* fix perceiver fp16

* hopefully fix tests

* add `accelerate` marker

* add to docs

* Update docs/source/en/testing.mdx

* uint8 -> bool

* fix copies

* style

* update test modeling commen when checking attention buffers

* style

* use logical not on random mask instead of subtraction with 1

* remove torch uint8

* quality

* remove modified modeling utils

* Update based on review
Co-authored-by: sgugger <sylvain.gugger@gmail.com>

---------
Co-authored-by: sgugger <sylvain.gugger@gmail.com>

* add pipeline

* update init

* add zero shot to init

* update inits and correct checkpoints

* update base to support input features

* add tests

* Update src/transformers/pipelines/zero_shot_audio_classification.py
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

* Update src/transformers/pipelines/zero_shot_audio_classification.py
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

* update pieline code

* use tiny checkpoint

* nits and expected value with tiny model

* style

* last nit on tests values

* fix styling

* fix collate fn that was casting t float

* update

---------
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

make concrete_args from outside available

* fix wrong url

* typos in english documentation

Fix docstring gpt2 config

* Override the decoding parameters of Seq2SeqTrainer

* Fix quality

* Fix max_length parameter

* Fix quality

* Remove redundant parameter max_length

* Separate the preprocess of train and validation to use different max_target_length

* Fix resume_from_checkpoint for deepspeed

Fix resume_from_checkpoint for deepspeed, by ensuring that the deepspeed engine is the one to load the checkpoint.

* Empty commit to trigger CI

* Removed deepspeed skipping 

Removed deepspeed skipping inside the _load_from_checkpoint function, as it is obsolete

* another adjustment

* Trigger CI

* trigger circleci

* style

---------
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>
Co-authored-by: Stas Bekman <stas@stason.org>

* fix gradient checkpointing bug

* fix gradient checkpointing bug

* ran make fix-copies

* fixed bug

* fixed bug

* updated expected

* prediction_length fix

* prediction_length default value

* default prediction_length 24

* revert back prediction_length default

* move prediction_length test

Generate - update cookie cutters to not initialize cache with training and gradient checkpointing (#21759)

* fix history

* input_features instead of input ids for TFWhisport doctest

* use translate intead of transcribe

* Return and rescale attention_mask

* Add SpecAugment to Whisper modeling

* Fix test

* Update docstring

* Add SpecAug related parameters to model config

* Add the _mask_input_features function to doc

* Fix quality

* Apply suggestions from code review
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Remove dev comments

* Add test

* Resolve conflict

* feat: mask {feature, time} prob fast tests

* Apply suggestions from code review
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

---------
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
Co-authored-by: sanchit-gandhi <sanchit@huggingface.co>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Different behavior in DistilBERT when using "inputs_embeds"
Fixes #21089

* fix failing test