GitLab

Skipping more tests

* config and tokenization(fast too) changed and ErnieEncoder added

* Slow Tokenization Added

* Tokenizer(slow) is now working and Fast Tokenizer removed

* Added Config code

* Added Base Model and utils

* ErnieMModel is now working

* All added except tests

* All tests passed except ErnieUIEM

* All tests passed

* all fixes done

* all fixes done

* fixed MAP

* fixed check_code_quality

* fixed Build PR Documentation issue

* Added changes(comments) and also updated to the latest upstream/main

* Added fixup

* Added # Copied comments

* Added fixup

* Added more comments and some nits

* Added fixup

* Fixed README_hd.md

* Added more fixes

* ErnieMTokenizer (being sentencepiece) protected and other docs edited

* Added code_quality fix

* Fixed for

* Added more fix

* modified AZ

* ernie-m tokenization test added!

* attention mask part fixed(with 0->self.config.pad_token_id)

* applied make fixup

* Skip high memory tests

* Skip high memory tests

* Remove unused import

* fix blip-2 ci

* fix blip-2 ci

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

* Fix typo in documentation.

* Remove trailing words typo in documentation.

* Add _add_sm_patterns_to_gitignore

* Add _is_world_process_zero() call and move patterns arg to constant

* Update git status time.sleep

* Apply make style

fix typo

Error (also in original) model, scaling only q matrix not qk.T dot product (qk.T/sqrt(dim_per_head)) (#21627)

* Error in model, scaling only q matrix not qK.T dot product (qk.T/sqrt(dim_per_head))

As per Vaswani et al, 2017 p.4
Is torch.matmul(q, k.transpose(2, 3)) / math.sqrt(dim_per_head) not q / math.sqrt(dim_per_head)
https://arxiv.org/pdf/1912.05372.pdf

Error was in original FlauBERT repo and effectively scales queries but not values
cf. https://github.com/getalp/Flaubert/pull/45/commits/6d176880ca3a1a8dfa2b76c97030bb51c5e917b8

* Update modeling_flaubert.py

Update to https://github.com/huggingface/transformers/pull/21627
make fixup
make repo_consistency

* Update modeling_xlm.py

* Update modeling_flaubert.py

* Update modeling_xlm.py

* Remove redundant computations, comb variable names

* Fix scores to cur_scores

FInal cleanup of TOKENIZER_FOR_DOC

* v1

* make fixup

* add more methods

* Add in big model inference to issue template

* Trigger

* Untrigger

* empty test commit

* fix env issue

* fix env issue

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

clarify available pipelines

* [deepspeed] performance docs

* fix

* re-org

* update

* update

* a new NCCL Collectives section

* inference

* Update docs/source/en/main_classes/deepspeed.mdx
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* suggestion

* Update docs/source/en/main_classes/deepspeed.mdx

* suggestion

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

* Update setup.py

* suggestions

* Translate quicktour to French

* Traduction missing task

* use fp16

* use fp16

* use fp16

* use fp16

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

Bump ipython in /examples/research_projects/decision_transformer

Bumps [ipython](https://github.com/ipython/ipython) from 8.1.1 to 8.10.0.
- [Release notes](https://github.com/ipython/ipython/releases)
- [Commits](https://github.com/ipython/ipython/compare/8.1.1...8.10.0

)

---
updated-dependencies:
- dependency-name: ipython
  dependency-type: direct:production
...
Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* annotated TFvisionEncoderDecoder input type hints
Co-authored-by: JuheonChu <chuj@dickinson.edu>
Co-authored-by: AdiaWu <wua@dickinson.edu>

* fixed failing tests

* make fix-copies

* failed test fix

* style fix

* revert

---------
Co-authored-by: JuheonChu <chuj@dickinson.edu>
Co-authored-by: AdiaWu <wua@dickinson.edu>
Co-authored-by: Matt <rocketknight1@gmail.com>

* fix bnb slow test

* make fixup

[MINOR] Fix link

I'm not sure this will also fix the currently broken link in the docs (Specifically here: https://huggingface.co/docs/transformers/model_doc/time_series_transformer) whereby clicking on `kashif` attempts to link to the following non-existent URL: https://huggingface.co/docs/transformers/model_doc/%3Chttps://huggingface.co/kashif

remove trailing word

skip test

* document question answering guide

* Added the list of supported models

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

* switched to AutoProcessor

* feedback addressed

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

* Update docs/source/en/tasks/document_question_answering.mdx
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>

* more feedback addressed

* addressed comments about evaluation loss

* added appropriate image link

* make style

* typo fix

* resolving toc conflict

* fixed the image link

---------
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>