GitLab

* Add initial BNB integration

* fixup! Add initial BNB integration

* Add bnb test decorator

* Update Adamw8bit option name

* Use the full bnb package name

* Overide bnb for all embedding layers

* Fix package name

* Formatting

* Remove unnecessary import

* Update src/transformers/trainer.py
Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>

* Rename AdamwBNB optimizer option

* Add training test checking that bnb memory utilization is lower

* fix merge

* fix merge; fix + extend new test

* cleanup

* expand bnb

* move all require_* candidates to testing_utils.py
Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>
Co-authored-by: Stas Bekman <stas@stason.org>

* Split file_utils in several submodules

* Fixes

* Add back more objects

* More fixes

* Who exactly decided to import that from there?

* Second suggestion to code with code review

* Revert wront move

* Fix imports

* Adapt all imports

* Adapt all imports everywhere

* Revert this import, will fix in a separate commit

* Fix failing GPU trainer tests

* Remove print statements

* [trainer] fix % 0

* sanity checks

* fix logging_strategy

* correction

* Update src/transformers/training_args.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* split the test into 4 sub-tests to avoid timeout

* fix decorator order

* bug fixes and a rename

* add extended DDP test

* ensure concurrent pytest workers use a unique port for torch.distributed.launch

* reword

* fixing tests

* cleanup

* added changes for uniformity

* modified files

* corrected typo

* fixed qa scripts

* fix typos

* fixed predict typo in qa no trainer

* fixed test file

* reverted trainer changes

* reverted trainer changes in custom exmaples

* updated readme

* added changes in deepspeed test

* added changes for predict and eval

* Base move

* Examples reorganization

* Update references

* Put back test data

* Move conftest

* More fixes

* Move test data to test fixtures

* Update path

* Apply suggestions from code review
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Address review comments and clean
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* relocate core integration tests

* add sys.path context manager

* cleanup

* try

* try2

* fix path

* doc

* style

* add dep

* add 2 more deps

* split seq2seq script, update docs

* needless diff

* fix readme

* remove test diff

* s/summarization/translation
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* cr

* fix arguments & better mbart/t5 refs

* copyright
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* reword readme
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* s/summarization/translation

* short script names

* fix tests

* fix isort, include mbart doc

* delete old script, update tests

* automate source prefix

* automate source prefix for translation

* s/translation/trans
Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>

* fix script name (short version)

* typos
Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>

* exact parameter
Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>

* remove superfluous source_prefix calls in docs

* rename scripts & warn for source prefix

* black

* flake8
Co-authored-by: theo <theo@matussie.re>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Suraj Patil <surajp815@gmail.com>
Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>

* Hotfix fairscale FSDP

* Evaluation works

* Save on process zero

* fix sharded ddp enum

* test fixes

* stronger validation + apex breaks other tests

* Ass support for ZeRO-2/3 and ZeRO-offload in fairscale

* Quality

* Rework from review comments

* Add doc

* Apply suggestions from code review
Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>

* Address review comments
Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>

* fix run_seq2seq.py; porting DeepSpeed tests to it

* unrefactor

* defensive programming

* defensive programming 2

* port the rest of the trainer tests

* style

* a cleaner scripts dir finder

* cleanup

* deepspeed bug fixes and tests

* manual wrap?

* Upstream (and rename) sortish sampler

* Use proper sampler

* Update src/transformers/trainer_pt_utils.py
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* deepspeed integration

* style

* add test

* ds wants to do its own backward

* fp16 assert

* Update src/transformers/training_args.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* style

* for clarity extract what args are being passed to deepspeed

* introduce the concept of self.wrapped_model

* s/self.wrapped_model/self.model_wrapped/

* complete transition to self.wrapped_model / self.model

* fix

* doc

* give ds its own init

* add custom overrides, handle bs correctly

* fix test

* clean up model_init logic, fix small bug

* complete fix

* collapse --deepspeed_config into --deepspeed

* style

* start adding doc notes

* style

* implement hf2ds optimizer and scheduler configuration remapping

* oops

* call get_num_training_steps absolutely when needed

* workaround broken auto-formatter

* deepspeed_config arg is no longer needed - fixed in deepspeed master

* use hf's fp16 args in config

* clean

* start on the docs

* rebase cleanup

* finish up --fp16

* clarify the supported stages

* big refactor thanks to discovering deepspeed.init_distributed

* cleanup

* revert fp16 part

* add checkpoint-support

* more init ds into integrations

* extend docs

* cleanup

* unfix docs

* clean up old code

* imports

* move docs

* fix logic

* make it clear which file it's referring to

* document nodes/gpus

* style

* wrong format

* style

* deepspeed handles gradient clipping

* easier to read

* major doc rewrite

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

* docs

* switch to AdamW optimizer

* style

* Apply suggestions from code review
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* clarify doc
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Add label smoothing in Trainer

* Add options for scheduler and Adafactor in Trainer

* Put Seq2SeqTrainer in the main lib

* Apply suggestions from code review
Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* Address review comments and adapt scripts

* Documentation

* Move test not using script to tests folder
Co-authored-by: Stas Bekman <stas00@users.noreply.github.com>
Co-authored-by: Patrick von Platen <patrick.v.platen@gmail.com>

* Reorganize example folder

* Continue reorganization

* Change requirements for tests

* Final cleanup

* Finish regroup with tests all passing

* Copyright

* Requirements and readme

* Make a full link for the documentation

* Address review comments

* Apply suggestions from code review
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Add symlink

* Reorg again

* Apply suggestions from code review
Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>

* Adapt title

* Update to new strucutre

* Remove test

* Update READMEs
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>

* fix DP case on multi-gpu

* make executable

* test all 3 modes

* use the correct check for distributed

* dp doesn't need a special case

* restore original name

* cleanup

* make generate work with multigpu

* better fix - thanks @sgugger

* Remove old deprecated arguments
Co-authored-by: LysandreJik <lysandre.debut@reseau.eseo.fr>

* Remove needless imports

* Fix tests
Co-authored-by: LysandreJik <lysandre.debut@reseau.eseo.fr>

Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

* move the helper code into testing_utils

* port test_trainer_distributed to work with pytest

* improve docs

* simplify notes

* doc

* doc

* style

* doc

* further improvements

* torch might not be available

* real fix

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

* make sure padding is implemented for non-padding tokens models as well

* add better error message

* add better warning

* remove results files

* Update examples/seq2seq/seq2seq_trainer.py

* remove unnecessary copy line

* correct usage of labels

* delete test files

* Make Seq2Seq Trainer more similar to Trainer

* fix typo

* fix seq2seq trainer

* remove from tests

* remove lock

* remove train files

* delete test files

* correct typo

* check at init

* make sure trainer is not slowed down on TPU

* correct isort

* remove use cache

* fix use cache

* add last use chache = false

* 2 beam output

* unassign/remove TODOs

* remove one more