GitLab

* Exclude Databricks from notebook env only if the runtime is below 11.0

* Dummy commit to trigger CI

* Empty commit to trigger CI

* Empty commit to trigger CI

* Empty commit to trigger CI

* Empty commit to trigger CI

* Empty commit to trigger CI

* Empty commit to trigger CI

* Empty commit to trigger CI

* Shifting labels for causal LM when using label smoother

When training CausalLM, loss is computed within model's foward() function and
labels are shifted internally. However, if label smoothing is applied, loss is
computed in trainer's compute_loss function and labels are not shifted.
This causes unintended confusion during the alignment of labels and corresponding
inputs. This commit is for resolving this confusion.

Resolves #17960

On branch shift_labels_for_causalLM
Changes to be committed:
	modified:   src/transformers/trainer.py
	modified:   src/transformers/trainer_pt_utils.py

* Update trainer.py

* Update src/transformers/trainer.py
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: ydshieh <ydshieh@users.noreply.github.com>

* Copy inputs to train and test step before modifying them, as this breaks things

* Add XLA tests, fix our loss functions to be XLA-compatible

* make fixup

* Update loss computation test to expect vector of per-sample losses

* Patch loss for TFLED

* Patch loss for TFAlbert

* Add a tf_legacy_loss config flag that enables old loss functions

* Stop using config.get() because it's not a dict

* Skip loss computation test for RAG because its loss is very strange and I'm afraid to rewrite it

* make fixup

* Add XLA-compatible RAG loss

* Fix dtype of loss mask for TFAlbert

* Fix test for XLNet too because it overrides the default one

* make fixup

* Fix config test

* No more depending on GPU NaN behaviour

* Add test, avoid potential zero division

* Fix test item assignment

* Fix loss computation masking test

* make fixup

* Fix dtype bugs

* [Flax] Add remat (gradient checkpointing)

* fix variable naming in test

* flip: checkpoint using a method

* fix naming

* fix class naming

* apply PVP's suggestions from code review

* make fix-copies

* fix big-bird, electra, roberta

* cookie-cutter

* fix flax big-bird

* move test to common

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

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

* add onnx support for BLOOM

* use TYPE_CHECKING for type annotations

* fix past_shape for bloom (different from gpt2)

* use logical_or instead of `+` for onnx support

* bigger `atol_for_validation` for larger bloom models

* copied -> taken because it's no longer an exact copy

* remove "copied from" comment
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* fixing fsdp autowrap functionality

* update version and quality

* update torch version to latest stable version

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

* skip some gpt_neox tests that require 80G RAM

* remove tests

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

* feat: add pipeline registry abstraction

- added `PipelineRegistry` abstraction
- updates `add_new_pipeline.mdx` (english docs) to reflect the api addition
- migrate `check_task` and `get_supported_tasks` from
  transformers/pipelines/__init__.py to
  transformers/pipelines/base.py#PipelineRegistry.{check_task,get_supported_tasks}
Signed-off-by: Aaron Pham <29749331+aarnphm@users.noreply.github.com>

* fix: update with upstream/main

chore: Apply suggestions from sgugger's code review
Signed-off-by: Aaron Pham <29749331+aarnphm@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* chore: PR updates

- revert src/transformers/dependency_versions_table.py from upstream/main
- updates pipeline registry to use global variables
Signed-off-by: Aaron Pham <29749331+aarnphm@users.noreply.github.com>

* tests: add tests for pipeline registry
Signed-off-by: Aaron Pham <29749331+aarnphm@users.noreply.github.com>

* tests: add test for output warning.
Signed-off-by: Aaron Pham <29749331+aarnphm@users.noreply.github.com>

* chore: fmt and cleanup unused imports
Signed-off-by: Aaron Pham <29749331+aarnphm@users.noreply.github.com>

* fix: change imports to top of the file and address comments
Signed-off-by: Aaron Pham <29749331+aarnphm@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* Add ONNX support for LayoutLMv3

* Update docstrings

* Update empty description in docstring

* Fix imports and type hints

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

* sharded conversion; add flag to control max hidden error

* better hidden name matching

* Add test: load TF from PT shards

* fix test (PT data must be local)

* trigger test failure

* upload revision poc

* Update src/transformers/pipelines/base.py
Co-authored-by: Julien Chaumond <julien@huggingface.co>

* up

* add test

* correct some stuff

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

* correct require flag
Co-authored-by: Julien Chaumond <julien@huggingface.co>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* doc: Unify training arg type annotations

* wip: extracting enum type from Union

* blackening

* Fix GPT-NeoX-20B past handling, swap attention computation to hopefully avoid NaN, update docs

* 20B tests

* first draft adding Flax-t5-encoder and Flax-mt5-encoder

* imports

* after make fixup

* flax t5 encoder test

* black on test

* make fix-copies

* clean

* all_model_classes -> tuple

* clean test

* is_encoder_decoder=False in t5-enc tester

* remove file docstring before FlaxT5Encoder

* black

* isort

* commit suggestions on src/transformers/models/t5/modeling_flax_t5.py
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* commit suggestions on src/transformers/models/t5/modeling_flax_t5.py
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* Apply suggestions from code review
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* remove _get_encoder_module

* self.decoder_seq_length -> self.encoder_seq_length as t5-enc does not have decoder

* bugfix - self.module_class is class itself, not instance;

* docs for mt5 and t5

* call -> __call__ in t5 doc

* FlaxMT5EncoderModel to TYPE_HINT

* run doc-builder to allow change the files
Co-authored-by: Suraj Patil <surajp815@gmail.com>

* Removed dead position_id code, fix #17893

* Removed unused var

* Now ignores removed (dead) dict key for backward comp

* add MobileViT

* fixup

* Update README.md
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* remove empty line
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* use clearer variable names

* rename to MobileViTTransformerLayer

* no longer inherit from nn.Sequential

* fixup

* fixup

* not sure why this got added twice

* rename organization for checkpoints

* fix it up

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

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

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

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

* Update tests/models/mobilevit/test_modeling_mobilevit.py
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

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

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

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

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

* code style improvements

* fixup

* Update docs/source/en/model_doc/mobilevit.mdx
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update docs/source/en/model_doc/mobilevit.mdx
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/models/mobilevit/configuration_mobilevit.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* Update src/transformers/models/mobilevit/configuration_mobilevit.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* download labels from hub

* rename layers

* rename more layers

* don't compute loss in separate function

* remove some nn.Sequential

* replace nn.Sequential with new MobileViTTransformer class

* replace nn.Sequential with MobileViTMobileNetLayer

* fix pruning since model structure changed

* fixup

* fix doc comment

* remove custom resize from feature extractor

* fix ONNX import

* add to doc tests

* use center_crop from image_utils

* move RGB->BGR flipping into image_utils

* fix broken tests

* wrong type hint

* small tweaks
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

* ExplicitEnum subclass str (JSON dump compatible)

* allow union if one of the types is str

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

* use explicit torch version
Co-authored-by: ydshieh <ydshieh@users.noreply.github.com>

* Fix all is_torch_tpu_available 

* Revert "Skip failing test until they are fixed."

This reverts commit 8f400775.

* Use `tiny-detr` checkpts from `hf-internal-testing`

* Add MVP model

* Update README

* Remove useless module

* Update docs

* Fix bugs in tokenizer

* Remove useless test

* Remove useless module

* Update vocab

* Remove specifying

* Remove specifying

* Add #Copied ... statement

* Update paper link

* Remove useless TFMvp

* Add #Copied ... statement

* Fix style in test mvp model

* Fix some typos

* Fix properties of unset special tokens in non verbose mode

* Update paper link

* Update MVP doc

* Update MVP doc

* Fix README

* Fix typos in docs

* Update docs

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

* chore: initial commit

Copied the torch implementation of regnets and porting the code to tf step by step. Also introduced an output layer which was needed for regnets.

* chore: porting the rest of the modules to tensorflow

did not change the documentation yet, yet to try the playground on the model

* Fix initilizations (#1)

* fix: code structure in few cases.

* fix: code structure to align tf models.

* fix: layer naming, bn layer still remains.

* chore: change default epsilon and momentum in bn.

* chore: styling nits.

* fix: cross-loading bn params.

* fix: regnet tf model, integration passing.

* add: tests for TF regnet.

* fix: code quality related issues.

* chore: added rest of the files.

* minor additions..

* fix: repo consistency.

* fix: regnet tf tests.

* chore: reorganize dummy_tf_objects for regnet.

* chore: remove checkpoint var.

* chore: remov unnecessary files.

* chore: run make style.

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

* chore: PR feedback I.

* fix: pt test. thanks to @ydshieh.

* New adaptive pooler (#3)

* feat: new adaptive pooler

Co-authored-by: @Rocketknight1

* chore: remove image_size argument.
Co-authored-by: matt <rocketknight1@gmail.com>
Co-authored-by: matt <rocketknight1@gmail.com>

* Empty-Commit

* chore: remove image_size comment.

* chore: remove playground_tf.py

* chore: minor changes related to spacing.

* chore: make style.

* Update src/transformers/models/regnet/modeling_tf_regnet.py
Co-authored-by: amyeroberts <aeroberts4444@gmail.com>

* Update src/transformers/models/regnet/modeling_tf_regnet.py
Co-authored-by: amyeroberts <aeroberts4444@gmail.com>

* chore: refactored __init__.

* chore: copied from -> taken from./g

* adaptive pool -> global avg pool, channel check.

* chore: move channel check to stem.

* pr comments - minor refactor and add regnets to doc tests.

* Update src/transformers/models/regnet/modeling_tf_regnet.py
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>

* minor fix in the xlayer.

* Empty-Commit

* chore: removed from_pt=True.
Co-authored-by: Sayak Paul <spsayakpaul@gmail.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: matt <rocketknight1@gmail.com>
Co-authored-by: amyeroberts <aeroberts4444@gmail.com>
Co-authored-by: NielsRogge <48327001+NielsRogge@users.noreply.github.com>