GitLab

* Revert "Add tie_weights() to LM heads and set bias in set_output_embeddings() (#28948)"

This reverts commit 725f4ad1.

* Revert "Patch to skip failing `test_save_load_low_cpu_mem_usage` tests (#29043)"

This reverts commit 4156f517.

* add add_dummy_prefix_space option to slow

* checking kwargs might be better. Should be there for all spm tokenizer IMO

* nits

* fix copies

* more copied

* nits

* add prefix space

* nit

* nits

* Update src/transformers/convert_slow_tokenizer.py

* fix inti

* revert wrong styling

* fix

* nits

* style

* updates

* make sure we use slow tokenizer for conversion instead of looking for the decoder

* support llama ast well

* update llama tokenizer fast

* nits

* nits nits nits

* update the doc

* update

* update to fix tests

* skip unrelated tailing test

* Update src/transformers/convert_slow_tokenizer.py

* add proper testing

* test decode as well

* more testing

* format

* fix llama test

* Apply suggestions from code review

* handle peft + compiled models

* add tests

* fixup

* adapt from suggestions

* clarify comment

* forgot to push the changes for 4bit ..

* trigger CI

* add RMSProp to Trainer

* revert some change

* Update src/transformers/trainer.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* generated text on A10G

* generated text in CI

* Apply suggestions from code review

add explanatory comments
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

---------
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

output_logits option behaves like output_scores, but returns the raw, unprocessed prediction logit scores,
ie. the values before they undergo logit processing and/or warping. The latter happens by default for the
regular output scores.

It's useful to have the unprocessed logit scores in certain circumstances. For example, unprocessed logit scores
are very useful with causallm models when one wants to determine the probability of a certain answer, e.g.
when asking a question with a yes/no answer. In that case getting the next-token probabilities of both "yes" and
"no" (and/or their relative ratio) is of interest for classification. The reason for getting these _before_ logit
processing and/or warping is b/c a) that can change the probabilities or b) reject the tokens of interest / reduce
the number of tokens to just 1.

For an example use-case see paper TabLLM: Few-shot Classification of Tabular Data with Large Language Models
by Stefan Hegselmann, Alejandro Buendia, Hunter Lang, Monica Agrawal, Xiaoyi Jiang, and David Sontag.
https://arxiv.org/abs/2210.10723



In addition:
- added dedicated unit test: tests/generation/test_utils/test_return_unprocessed_logit_scores
  which tests return of logics with output_logits=True in generation.
- set output_logits=True in all other generation unit tests, that also have output_scores=True.

Implemented @gante's and @amyeroberts review feedback
Co-authored-by: kx79wq <max.baak@ing.com>

Fix test

Update test_mixed_int8.py

* Add chat support to text generation pipeline

* Better handling of single elements

* Deprecate ConversationalPipeline

* stash commit

* Add missing add_special_tokens kwarg

* Update chat templating docs to refer to TextGenerationPipeline instead of ConversationalPipeline

* Add ✨TF✨

 tests

* @require_tf

* Add type hint

* Add specific deprecation version

* Remove unnecessary do_sample

* Remove todo - the discrepancy has been resolved

* Update src/transformers/tokenization_utils_base.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update src/transformers/pipelines/text_generation.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* FIx trainer test

* Update tests/trainer/test_trainer.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* pass through trust_remote_code for dynamically loading unregistered tokenizers specified by config
add test

* change directories back to previous directory after test

* fix ruff check

* Add a note to that block for future in case we want to remove it later

---------
Co-authored-by: Matt <rocketknight1@gmail.com>

* fix heuristic num_assistant_tokens_schedule

* Update src/transformers/generation/configuration_utils.py
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* Update src/transformers/generation/candidate_generator.py
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* Update utils.py

check that candidate_generator.assistant_model exists since some some speculations (like ngram and PLD) don't have assistant_model attribute

* Update src/transformers/generation/candidate_generator.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Update tests/generation/test_utils.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* make fixup

* merge conflict

* fix docstring

* make fixup

---------
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* fix max_length for inputs_embeds

* make style

* Update src/transformers/generation/utils.py
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* Static Cache: load models with MQA or GQA (#28975)

* fix

* fix tests

* fix tests

* Update src/transformers/generation/utils.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* more fixes

* make style

---------
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Script & Manual edition

* Update

* Patch to skip currently failing tests

* Whoops - wrong place

* Update ms_deform_attn_cuda.cu

* Update ms_deform_attn_cuda.cuh

* Update modeling_deformable_detr.py

* Update src/transformers/models/deformable_detr/modeling_deformable_detr.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Update modeling_deformable_detr.py

* python utils/check_copies.py --fix_and_overwrite

* Fix dtype missmatch error

* Update test_modeling_deformable_detr.py

* Update test_modeling_deformable_detr.py

* Update modeling_deformable_detr.py

* Update modeling_deformable_detr.py

---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* wow I was scared!

* fix everything

* nits

* make it BC?

* add todo

* nits

* is_tracing should still be used to pass tracing tests

* nits

* some nits to make sure genration works with static cache uncompiled

* fix sdpa

* fix FA2 for both static and dynamic in a better way?

* style

* fix-copies

* fix fix copies

* fix sequential beam searcg

* style

* use `keys_to_ignore`

* nit

* correct dtype inference when init

* :( the fix for FA2 is still not optimal to investigate!

* styling

* nits

* nit

* this might work better

* add comment

* Update src/transformers/models/llama/modeling_llama.py

* "position_ids" -> "cache_position"

* style

* nit

* Remove changes that should no be propagatted just yet

* Apply suggestions from code review

* Styling

* make sure we raise an errir for static cache with FA2 enabled

* move  to the bottom of the signature

* style

* Update src/transformers/models/llama/modeling_llama.py
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

* Update src/transformers/models/llama/modeling_llama.py

* nit in the name

---------
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

FIX [`Trainer` / tags]: Fix trainer + tags when users do not pass `"tags"` to `trainer.push_to_hub()` (#29009)

* fix trainer tags

* add test

* Enable instantiating model with pretrained backbone weights

* Clarify pretrained import

* Use load_backbone instead

* Add backbone_kwargs to config

* Pass kwargs to constructors

* Fix up

* Input verification

* Add tests

* Tidy up

* Update tests/utils/test_backbone_utils.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

---------
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Add tie_weights() to LM heads and set bias in set_output_embeddings()

The bias were not tied correctly in some LM heads, and this change should fix that.

* Moving test_save_and_load_low_cpu_mem_usage to ModelTesterMixin

* Adding _tie_weights() to MPNet and Vilt

* Skip test for low cpu mem usage for Deta/DeformableDetr since they cannot init on meta device

* Rename to test name to save_load to match the convention

* Introduce acceleratorconfig dataclass

* Extra second warn

* Move import

* Try moving import under is_accelerate_available

* Quality

* Apply suggestions from code review
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Clean

* Remove to_kwargs

* Change version

* Improve tests by including dispatch and split batches

* Improve reliability

* Update tests/trainer/test_trainer.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* Fixup tests and review nits

* Make tests pass

* protect import

* Protect import

* Empty-Commit

* Make training_args.to_dict handle the AcceleratorConfig

---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

Co-authored-by: unit_test <test@unit.com>

* aqlm init

* calibration and dtypes

* docs

* Readme update

* is_aqlm_available

* Simpler link in docs

* Test TODO real reference

* init _import_structure fix

* AqlmConfig autodoc

* integration aqlm

* integrations in tests

* docstring fix

* legacy typing

* Less typings

* More kernels information

* Performance -> Accuracy

* correct tests

* remoced multi-gpu test

* Update docs/source/en/quantization.md
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>

* Update src/transformers/utils/quantization_config.py
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* Brought back multi-gpu tests

* Update src/transformers/integrations/aqlm.py
Co-authored-by: Marc Sun <57196510+SunMarc@users.noreply.github.com>

* Update tests/quantization/aqlm_integration/test_aqlm.py
Co-authored-by: Marc Sun <57196510+SunMarc@users.noreply.github.com>

---------
Co-authored-by: Andrei Panferov <blacksamorez@yandex-team.ru>
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>
Co-authored-by: Marc Sun <57196510+SunMarc@users.noreply.github.com>

* First draft

* Add CLIPForImageClassification

* Remove scripts

* Fix doctests

* Add `StableLM`

* fix(model): re-create from `huggingface-cli add-new-model-like persimmon`

* fix: re-add changes to address comments

* fix(readme): add links to paper

* fix(tokenization_auto): remove `GPTNeoXTokenizerFastFast` ref

* fix(tests): re-add `@slow` decorator to integration tests

* fix(tests): import slow...

* fix(readme_hd): remove whitespace edit

* fix(tokenizer): auto tokenizer tuple

* skip doctests for `modeling_stablelm`

* Update the processing so bbox coords are adjusted for padding

* Just pad masks

* Tidy up, add tests

* Better tests

* Fix yolos and mark as slow for pycocotols

* Fix yolos - return_tensors

* Clarify padding and normalization behaviour

* add sudachi_projection option

* Upgrade sudachipy>=0.6.8

* add a test case for sudachi_projection

* Compatible with older versions of SudachiPy

* make fixup

* make style

* error message for unidic download

* revert jumanpp test cases

* format options for sudachi_projection
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* format options for sudachi_split_mode and sudachi_dict_type

* comment

* add tests for full_tokenizer kwargs

* pass projection arg directly

* require_sudachi_projection

* make style

* revert upgrade sudachipy

* check is_sudachi_projection_available()

* revert dependency_version_table and bugfix

* style format

* simply raise ImportError
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* simply raise ImportError

---------
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* refactor with addedtokens decoder

* style

* get rid of lang code to id

* style

* keep some things for BC

* update tests

* add the mask token at the end of the vocab

* nits

* nits

* fix final tests

* style

* nits

* Update src/transformers/models/nllb/tokenization_nllb_fast.py
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

* nits

* style?

* Update src/transformers/convert_slow_tokenizer.py

* make it a tad bit more custom

* ruff please stop
Co-Authored by avidale

<dale.david@mail.ru>

* Update
Co-authored-by: avidale <dale.david@mail.ru>

* Update
Co-authored-by: avidale <dale.david@mail.ru>

* oupts

* ouft

* nites

* test

* fix the remaining failing tests

* style

* fix failing test

* ficx other test

* temp dir + test the raw init

* update test

* style

---------
Co-authored-by: amyeroberts <22614925+amyeroberts@users.noreply.github.com>

Add language identifiers to code blocks

* Fix pipeline

* Remove print statements

* Address comments

* Address issue

* Remove unused imports

* support batched input for decoder start ids

* Fix typos
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* minor changes

* fix: decoder_start_id as list

* empty commit

* empty commit

* empty commit

* empty commit

* empty commit

* empty commit

* empty commit

* empty commit

* empty commit

---------
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

Co-authored-by: fxmarty <9808326+fxmarty@users.noreply.github.com>
Co-authored-by: Younes Belkada <49240599+younesbelkada@users.noreply.github.com>
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* change warning to exception

* Update src/transformers/generation/utils.py
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>

* validate `max_new_tokens` > 0 in `GenerationConfig`

* fix truncation test parameterization in `TextGenerationPipelineTests`

---------
Co-authored-by: Joao Gante <joaofranciscocardosogante@gmail.com>