GitLab

* Per-folder tests reorganization
Co-authored-by: sgugger <sylvain.gugger@gmail.com>
Co-authored-by: Stas Bekman <stas@stason.org>

In the code, `do_normalize` defaults to True

* Enabling Beit SegFormer to `image-segmentation`.

* Fixing the score.

* Fix import ?

* Missing in type hint.

* Multiple test fixes:

- Add `raw_image` support. It should be the default IMHO since in Python
  world it doesn't make any sense to base64 encode the image (Sorry
  @mishig, didn't catch that in my review). I really think we should
  consider breaking BC here.
- Add support for Segformer tiny test (needed
  `SegformerModelTester.get_config` to enable TinyConfig
  @NielsRogge)
- Add the check that `batch_size` works correctly on that pipeline.
  Uncovered that it doesn't for Detr, which IMO is OK since images
  after `feature_extractor` don't have the same size. Comment should
  explain.

* Type hint as a string.

* Make fixup + update black.

* torch+vision protections.

* Don't use torchvision, use F.interpolate instead (no new dep).

* Last fixes for Segformer.

* Update test to reflect new image (which was broken)

* Update tests.

* Major BC modification:

- Removed the string compressed PNG string, that's a job for users
`transformers` stays in python land.
- Removed the `score` for semantic segmentation. It has hardly a meaning
  on its own in this context.
- Don't include the grayscale with logits for now (which could enable
  users to get a sense of confidence). Might be done later.
- Don't include the surface of the mask (could be used for sorting by
  users, to filter out small masks). It's already calculable, and
  it's easier to add later, than to add now and break later if we need.

* `make fixup`.

* Small changes.

* Rebase + doc fixup.

* [ViLT] Fix checkpoint url in config

* Apply suggestions from code review
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

(`hf-internal-testing/tiny-clip` for instance)

* [Proposal] Adding ZeroShotImageClassificationPipeline

- Based on CLIP

* WIP, Resurection in progress.

* Resurrection... achieved.

* Reword handling different `padding_value` for `feature_extractor` and
`tokenizer`.

* Thanks doc-builder !

* Adding docs + global namespace `ZeroShotImageClassificationPipeline`.

* Fixing templates.

* Make the test pass and be robust to floating error.

* Adressing suraj's comments on docs mostly.

* Tf support start.

* TF support.

* Update src/transformers/pipelines/zero_shot_image_classification.py
Co-authored-by: Suraj Patil <surajp815@gmail.com>
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* Fix `HfArgumentParser` when passing a generator

* Add missing import

* Always convert `dataclass_types` into a list

* [Wav2Vec2 Time Stamps]

* Add first version

* add word time stamps

* Fix

* save intermediate space

* improve

* [Finish CTC Tokenizer]

* remove @

* remove @

* push

* continue with phonemes

* up

* finish PR

* up

* add example

* rename

* finish

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

* correct split

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

* Add GeLU10 (clipped version of GeLU) to transformers to improve quantization performances.

* Add unittests.

* Import tensorflow after `is_tf_available` check.

* Fix tensorflow wrong function `tf.tensor` to `tf.constant`

* style.

* use `tf.math.max`

* Fix tf tests.

* style.

* style style style style style style

* style style style style style style

* Address @sgugger comments.

* Fix wrong operator for raising ValueError for ClippedGELUActivation.

* TF train_step docstring

* Working example with to_tf_dataset

* updated text_classification

* more comments

* Add layer_idx to CrossAttention

* Add layer_idx to crossattention of ImageGPT model

* fix bug in PT speech-encoder-decoder

* add pt test for `inputs is not None`

* fix test

* new pt test

* Update tests/test_modeling_speech_encoder_decoder.py

* make fixup
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

`HfDeepSpeedConfig` accepts a dictionary or path to `.json` file containing DS configurations, not `TrainingArguments`.

* Init PLBART

* Add missing configuration file

* Add conversion script and configurationf ile

* Fix style

* Update modeling and conversion scripts

* Fix scale embedding in config

* Add comment

* Fix conversion script

* Add classification option to conversion script

* Fix vocab size in config doc

* Add tokenizer files from MBart50

* Allow no lang code in regular tokenizer

* Add PLBart Tokenizer Converters

* Remove mask from multi tokenizer

* Remove mask from multi tokenizer

* Change from MBart-50 to MBart tokenizer

* Fix names and modify src/tgt behavior

* Fix imports for tokenizer

* Remove <mask> from multi tokenizer

* Fix style

* Change tokenizer_class to processor_class

* Add attribute map to config class

* Update modeling file to modified MBart code

* Update configuration file to MBart style configuration

* Fix tokenizer

* Separate tokenizers

* Fix error in tokenization auto

* Copy MBart tests

* Replace with MBart tokenization tests

* Fix style

* Fix language code in multi tokenizer

* Fix configuration docs

* Add entry for plbart_multi in transformers init

* Add dummy objects and fix imports

* Fix modeling tests

* Add TODO in config

* Fix copyright year

* Fix modeling docs and test

* Fix some tokenization tests and style

* Add changes from review

* Fix copies

* Fix docs

* Fix docs

* Fix style

* Fix year

* Add changes from review

* Remove extra changes

* Fix base tokenizer and doc

* Fix style

* Fix modeling and slow tokenizer tests

* Remove Multi-tokenizer Converter and Tests

* Delete QA model and Multi Tokenizer dummy objects

* Fix repo consistency and code quality issues

* Fix example documentation

* Fix style

* Remove PLBartTokenizer from type checking in init

* Fix consistency issue

* Add changes from review

* Fix style

* Remove PLBartTokenizerFast

* Remove FastTokenizer converter

* Fix AutoTokenzier mapping

* Add plbart to toctree and fix consistency issues

* Add language codes tokenizer test

* Fix styling and doc issues

* Add fixes for failing tests

* Fix copies

* Fix failing modeling test

* Change assert to assertTrue in modeling tests

* add undo padding

* fix

* fix tuple issue

* make style and quality

* move unpad logic to LongformerEncoder + unpad attentions + update tests

* move unpad logic to TFLongformerEncoder
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

Very big changes concerning the tokenizer fast of CLIP which did not correspond to the tokenizer slow of CLIP
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Add first draft

* Make model importable

* Make SwinForMaskedImageModeling importable

* Fix imports

* Add missing inits

* Add support for Swin

* Fix bug

* Fix bug

* Fix another bug

* Fix Swin MIM implementation

* Fix default encoder stride

* Fix Swin

* Add print statements for debugging

* Add image_size data argument

* Fix Swin

* Fix image_size

* Add print statements for debugging

* Fix print statement

* Remove print statements

* Improve reshaping of bool_masked_pos

* Add support for DeiT, fix tests

* Improve docstrings

* Apply new black version

* Improve script

* Fix bug

* Improve README

* Apply suggestions from code review

* Remove DS_Store and add to gitignore

* Apply suggestions from code review + fix BEiT Flax

* Revert BEiT changes

* Improve README

* Fix code quality

* Improve README
Co-authored-by: Niels Rogge <nielsrogge@Nielss-MBP.localdomain>
Co-authored-by: Niels Rogge <nielsrogge@Nielss-MacBook-Pro.local>

* Added all files, PoolFormerFeatureExtractor still failing tests

* Fixed PoolFormerFeatureExtractor not being able to import

* Completed Poolformer doc

* Applied Suggested fixes

* Fixed errors in modeling_auto.py

* Fix feature extractor, convert docs to Markdown, styling of code

* Remove PoolFormer from check_repo and fix integration test

* Remove Poolformer from check_repo

* Fixed configuration_poolformer.py docs and removed inference.py from poolformer

* Ran with black v22

* Added PoolFormer to _toctree.yml

* Updated poolformer doc

* Applied suggested fixes and added on README.md

* Did make fixup and make fix-copies, tests should pass now

* Changed PoolFormer weights conversion script name and fixed README

* Applied fixes in test_modeling_poolformer.py and modeling_poolformer.py

* Added PoolFormerFeatureExtractor to AutoFeatureExtractor API
Co-authored-by: Niels Rogge <nielsrogge@Nielss-MBP.localdomain>

* Implement activations as pytorch modules

* Apply fixup

* Add missing tests for activations

* Update docstring
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* fix doc

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

* Add push_to_hub method to processors

* Fix test

* The other one too!

* Add register method to AutoProcessor

* Update src/transformers/models/auto/processing_auto.py
Co-authored-by: Lysandre Debut <lysandre.debut@reseau.eseo.fr>
Co-authored-by: Lysandre Debut <lysandre.debut@reseau.eseo.fr>

* fix attn

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

* Add push_to_hub method to processors

* Fix test

* The other one too!

* [t5] faster/leaner custom layer norm

* wip

* apex.normalization.FusedRMSNorm

* cleanup

* cleanup

* add doc

* add catch all

* Trigger CI

* expand

* fix TFSequenceSummary

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

* [pipeline doc] fix api

* remove duplicate

* TF generate start refactor

* Add tf tests for sample generate

* re-organize

* boom boom

* Apply suggestions from code review

* re-add

* add all code

* make random greedy pass

* make encoder-decoder random work

* further improvements

* delete bogus file

* make gpt2 and t5 tests work

* finish logits tests

* correct logits processors

* correct past / encoder_outputs drama

* refactor some methods

* another fix

* refactor shape_list

* fix more shape list

* import shape
_list

* finish docs

* fix imports

* make style

* correct tf utils

* Fix TFRag as well

* Apply Lysandre's and Sylvais suggestions

* Update tests/test_generation_tf_logits_process.py
Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* Update src/transformers/tf_utils.py
Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

* remove cpu according to gante

* correct logit processor
Co-authored-by: Matt <Rocketknight1@users.noreply.github.com>

Co-authored-by: Giuseppe Attanasio <giuseppeattanasio6@gmail.com>
Co-authored-by: Giuseppe Attanasio <giuseppeattanasio6@gmail.com>

* Re-export `KeyDataset`.

* Update the docs locations.

* Improve the parameter `bad_word_ids' usage

* Update the bad_words_ids strategy