GitLab

* [WIP] Add FLAVA model

This PR aims to add [FLAVA](ihttps://arxiv.org/abs/2112.04482) model to the transformers repo.

Following checklist delineates the list of things to be done for this PR
to be complete:

[x] Flava init
[x] Flava base models
[x] Flava layers
[x] Flava Configs
[x] Flava encoders
[x] Flava pretraining models
[ ] Flava classification/retrieval models (To be added in a separate PR)
[x] Documentation updates 
[x] Imports updates 
[x] Argstring updates
[x] Flava pretrained checkpoints 
[x] Flava tests
[x] Flava processors 
[x] Sanity check
[x] Lint

* Ensure tensors are at least 1d for pad and concat

* Compatibility

* Fix

* Fix

* Add test

* Retrigger CI

* Consistency with master

* Retrigger CI

* Fix markdown code block

* Use consistent spelling for self-attention

* Fix typos and phrasing

* Fix code style

* Remove unneeded columns for IterableDataset

* Add test

* Update trainer tests

* Edit docstring

* Lint

* Apply feedback

* Apply feedback

* add seed worker and set_deterministic_seed_for_cuda function to enforce reproducability

* change function name to enable determinism, add docstrings, reproducability support for tf

* change function name to enable_determinism_for_distributed_training

* revert changes in set_seed and call set_seed within enable_full_determinism

* add one position argument for seed_worker function

* add full_determinism flag in training args and call enable_full_determinism when it is true

* add enable_full_determinism to documentation

* apply make fixup after the last commit

* 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>

* Create RetriBERT tests folder

* Add missing RetriBERT tokenizer test file

* Apply style corrections

* Add non-english filter

* Update tests/retribert/test_tokenization_retribert.py
Co-authored-by: SaulLu <55560583+SaulLu@users.noreply.github.com>

* Update tests/retribert/test_tokenization_retribert.py
Co-authored-by: SaulLu <55560583+SaulLu@users.noreply.github.com>

* Move test files to new directory

* Update import path for testing utils to new test file structure
Co-authored-by: SaulLu <55560583+SaulLu@users.noreply.github.com>

Co-authored-by: kuanwee.heng <kuanwee.heng@aaqua.live>

* propagate attention_mask dtype

* fixup&style

* add torch SGD and Adagrad optimizer bits

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

* remove duplicate example

* remove code block

* unhardcode pretrained model path, make it a class var

* add tests for mobilebert tokenizer

* allow tempfiles for vocab & merge similarity test to autodelete

* add explanatory comments

* remove unused imports, let make style do its.. thing

* remove inheritance and use BERT tok tests for MobileBERT

* Update tests/mobilebert/test_tokenization_mobilebert.py
Co-authored-by: SaulLu <55560583+SaulLu@users.noreply.github.com>

* amend class names, remove unused import, add fix for mobilebert's hub pathname

* unhardcode pretrained model path, make it a class var

* add tests for mobilebert tokenizer

* allow tempfiles for vocab & merge similarity test to autodelete

* add explanatory comments

* remove unused imports, let make style do its.. thing

* remove inheritance and use BERT tok tests for MobileBERT

* Update tests/mobilebert/test_tokenization_mobilebert.py
Co-authored-by: SaulLu <55560583+SaulLu@users.noreply.github.com>

* amend class names, remove unused import, add fix for mobilebert's hub pathname

* amend paths for model tests being in models/ subdir of /tests

* explicitly rm test from prev path
Co-authored-by: SaulLu <55560583+SaulLu@users.noreply.github.com>

* add support for MLFLOW_FLATTEN_PARAMS

* ensure key is str

* fix style and update warning msg

* Empty commit to trigger CI

* fix bug in check_inits.py

* add unittest for flatten_dict utils

* fix 'NoneType' object is not callable on __del__

* add generic flatten_dict unittest to SPECIAL_MODULE_TO_TEST_MAP

* fix style

* model zoo take 2

* add deberta

* new param for zero2

* doc update

* doc update

* add layoutlm

* bump deepspeed

* add deberta-v2, funnel, longformer

* new models

* style

* add t5_v1

* update TAPAS status

* reorg problematic models

* move doc to another PR

* style

* fix checkpoint check test

* making progress on more models running

* cleanup

* new version

* cleanup

* [trainer] sharded _load_best_model

probably needs a test?

* undo delete

Co-authored-by: Dom Miketa <dmiketa@exscientia.co.uk>

* LogSumExp trick `question_answering` pipeline.

* Adding a failing test.

* ensure mlflow.end_run() is executed at end of training when mlflow.start_run() was executed by the callback

* add debug msg

* add support for MLFLOW_TAGS, MLFLOW_RUN_ID, and MLFLOW_NESTED_RUN

* update to support python 3.6+

* Validate env variables using ENV_VARS_TRUE_VALUES

* Empty-Commit

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

- Adds auto_batch_size finder 
- Moves training loop to an inner training loop

* rebase and isort

* modify cookiecutter init

* fix cookiecutter auto imports

* fix clean_frameworks_in_init

* fix add_model_to_main_init

* blackify

* replace unnecessary f-strings

* update yolos imports

* fix roberta import bug

* fix yolos missing dependency

* fix add_model_like and cookiecutter bug

* fix repository consistency error

* modify cookiecutter, fix add_new_model_like

* remove stale line
Co-authored-by: Dom Miketa <dmiketa@exscientia.co.uk>

* PyTorch FSDP integration in Trainer

* reformatting

make style and make quality are now compliant.

* Updating dependency check

* Trigger CI
Co-authored-by: Sylvain Gugger <Sylvain.gugger@gmail.com>

* update docs of length_penalty

* Revert "update docs of length_penalty"

This reverts commit 466bf4800b75ec29bd2ff75bad8e8973bd98d01c.

* add mobilebert onnx config

* address suggestions

* Update auto.mdx

* Update __init__.py

* Update features.py

* Add type hints for remaining BigBirdPegasus models

Here I added type hints to the BigBirdPegasusForCausalLM class.

* Add missing type hints for Data2VecText models

Added type hints to the Data2VecTextForCausalLM, Data2VecTextForMaskedLM,
Data2VecTextForMultipleChoice, Data2VecTextForQuestionAnswering,
Data2VecTextForSequenceClassification, and
Data2VecTextForTokenClassification classes.

LayoutLMv2Processor: ensure 1-to-1 mapping between images and samples in case of overflowing tokens (#17092)

* add get_overflowing_images function to ensure 1-to-1 mapping between samples and images in LayoutLMv2Processor

* make style

* add test for overflowing_tokens, change assert to ValueError, avoiding unrelated formatting changes

* change line length by passing --preview into black

* split single_gpu and multi_gpu

* update needs in send_result
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

* Add onnx configuration for bigbird-pegasus

* Modify docs

* fix report cat path

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

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

* Fix use of mlflow.active_run() and add proper support for MLFLOW_EXPERIMENT_NAME

* Fix code style (make style)

Added type hints for the BERTGenerationEncoder and BERTGenerationDecoder
classes.