Forbid `PretrainedConfig` from saving `generate` parameters; Update deprecations in `generate`-related code 🧹  (#32659)

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

* Add .float() in all generation methods logit outputs

* Switch float-casting of logits to training only for main models

* Add `num_logits_to_keep` in Llama and add it by default in generate

* Apply style

* Add num_logits_to_keep as arg in prepare_input_for_generation

* Add support for Mistral

* Revert models except llama and mistral

* Fix default None value in _supports_num_logits_to_keep()

* Fix dimension of dummy input

* Add exception for prophetnet in _supports_num_logits_to_keep()

* Update _supports_num_logits_to_keep() to use inspect.signature()

* Add deprecation cycle + remove modification with pretraining_tp

* Apply style

* Add most used models

* Apply style

* Make `num_logits_to_keep` an int in all cases to remove if-else clause

* Add compile check for the warning

* Fix torch versions

* style

* Add gemma2

* Update warning version

* Add comment about .float operations in generation utils

* Add tests in GenerationTesterMixin and ModelTesterMixin

* Fix batch size for assisted decoding in tests

* fix small issues in test

* refacor test

* fix slicing removing dim issue

* Add nemotron support (should fix check-copy issue in CIs)

* Trigger new CIs

* Trigger new CIs

* Bump version

* Bump version in TODO

* Trigger CIs

* remove blank space

* Trigger CIs

fix outdated link

* docs: ko: tasks/knowledge_distillation_for_image_classification.md

* feat: nmt draft

* fix: manual edits

* Apply suggestions from code review

Co-authored-by: Chulhwa (Evan) Han <cjfghk5697@ajou.ac.kr>

* Apply suggestions from code review

Co-authored-by: Chulhwa (Evan) Han <cjfghk5697@ajou.ac.kr>

* Apply suggestions from code review

Co-authored-by: Ahnjj_DEV <ahnjj.dev@gmail.com>

* Apply suggestions from code review

Co-authored-by: Ahnjj_DEV <ahnjj.dev@gmail.com>

* Apply suggestions from code review

Co-authored-by: Ahnjj_DEV <ahnjj.dev@gmail.com>

* Apply suggestions from code review

Co-authored-by: Chulhwa (Evan) Han <cjfghk5697@ajou.ac.kr>

* Apply suggestions from code review

Co-authored-by: Chulhwa (Evan) Han <cjfghk5697@ajou.ac.kr>

* Apply suggestions from code review

Co-authored-by: Chulhwa (Evan) Han <cjfghk5697@ajou.ac.kr>

* Apply suggestions from code review

* Apply suggestions from code review

* Apply suggestions from code review

* Apply suggestions from code review

---------

Co-authored-by: Chulhwa (Evan) Han <cjfghk5697@ajou.ac.kr>
Co-authored-by: Ahnjj_DEV <ahnjj.dev@gmail.com>

fix save_pretrained

fix: (issue #32689) `AttributeError` raised when using `Trainer` with `eval_on_start=True` in Jupyter Notebook. (#32849)

fix: `AttributeError` raised when using `Trainer` with `eval_on_start=True` in Jupyter Notebook.

* add chat_template to gguf tokenizer

* add template through tokenizer config

Do not call torch.repeat_interleave if expand_size is 1

* fix

* >= 0.3.0

---------

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

* Update hub.py

* Update errors

* Apply suggestions from code review

Co-authored-by: Lucain <lucainp@gmail.com>

---------

Co-authored-by: Amy Roberts <22614925+amyeroberts@users.noreply.github.com>
Co-authored-by: Lucain <lucainp@gmail.com>

* separate step to download nltk files

* duplicated

* rm comma

* add 4bit optimizer

* style

* fix msg

* style

* add qgalore

* Revert "add qgalore"

This reverts commit 25278e805f24d5d48eaa0638abb48de1b783a3fb.

* style

* version check

Co-authored-by: Gal Cohen <galc@ai21.com>

Added missing huggingface_hub installation to workflows.

* try test updates

* a few more changes

* a few more changes

* a few more changes

* [run slow] jamba

* skip logits checks on older gpus

* [run slow] jamba

* oops

* [run slow] jamba

* Update tests/models/jamba/test_modeling_jamba.py

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

* Update tests/models/jamba/test_modeling_jamba.py

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

---------

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

commit

fix: [whisper] don't overwrite GenerationConfig's `return_timestamps` when `return_timestamps` is not passed to `generate` function (#31296)

[whisper] don't overwrite return_timestamps when not passed to generate

* Update README.md

* Update README.md

* Add README_ar.md to i18n/README_de.md

* Add README_ar.md to i18n/README_es.md

* Add README_ar.md to i18n/README_fr.md

* Add README_ar.md to i18n/README_hd.md

* Add README_ar.md to i18n/README_ja.md

* Add README_ar.md to i18n/README_ko.md

* Add README_ar.md to i18n/README_pt-br.md

* Add README_ar.md to i18n/README_ru.md

* Add README_ar.md to i18n/README_te.md

* Add README_ar.md to i18n/README_vi.md

* Add README_ar.md to i18n/README_vi.md

* Add README_ar.md to i18n/README_zh-hans.md

* Add README_ar.md to i18n/README_zh-hant.md

* Create README_ar.md

* link for optimizer names

Add a note and link to where the user can find more optimizer names easily because there are many more optimizers than are mentioned in the docstring.

* make fixup

* Replace .norm() with decomposed version for executorch export

* [run_slow] clip

Bump nltk in /examples/research_projects/decision_transformer

Bumps [nltk](https://github.com/nltk/nltk) from 3.7 to 3.9.
- [Changelog](https://github.com/nltk/nltk/blob/develop/ChangeLog)
- [Commits](https://github.com/nltk/nltk/compare/3.7...3.9

)

---
updated-dependencies:
- dependency-name: nltk
  dependency-type: direct:production
...

Signed-off-by: dependabot[bot] <support@github.com>
Co-authored-by: dependabot[bot] <49699333+dependabot[bot]@users.noreply.github.com>

* mamba2 uses norm_before_gate=False

* small nit

* remove norm_before_gate flag and follow False path only

Co-authored-by: Gal Cohen <galc@ai21.com>

add nx

* Update min version of accelerate to 0.26.0

* dev-ci

* update min version in import

* remove useless check

* dev-ci

* style

* dev-ci

* dev-ci

* support head dim

* fix the doc

* fixup

* add oproj

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

* update

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

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

* Update

Co-authored-by: Yoshi Suhara <suhara@users.noreply.github.com>

---------

Co-authored-by: bzantium <bzantium@users.noreply.github.com>
Co-authored-by: Yoshi Suhara <suhara@users.noreply.github.com>

Fixed whisper-large-v2 model link in docs.

* fix cache when using input embeddings

* simplify check, we can always add input ids seq len since its 0 in first pass

* fix mamba left padding

* Apply suggestions from code review

Co-authored-by: Pablo Montalvo <39954772+molbap@users.noreply.github.com>

* fix copies

* test with `inputs_embeds`

* Update src/transformers/models/falcon_mamba/modeling_falcon_mamba.py

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

* copies

* clairfy

* fix last comments

* remove

---------

Co-authored-by: Pablo Montalvo <39954772+molbap@users.noreply.github.com>
Co-authored-by: Arthur <48595927+ArthurZucker@users.noreply.github.com>

* fix gguf config vocab size

* minor fix

* link issue

* fix: Parameterized norm freezing

For the R18 model, the authors don't freeze norms in the backbone.

* Update src/transformers/models/rt_detr/configuration_rt_detr.py

Co-authored-by: Pavel Iakubovskii <qubvel@gmail.com>

---------

Co-authored-by: Pavel Iakubovskii <qubvel@gmail.com>

* Support save/load ckpt for XLA FSDP

* Fix bug for save

* Fix style

* reserve sharded ckpt and better file naming

* minor fix

Co-authored-by: Zach Mueller <muellerzr@gmail.com>

* add is_fsdp_xla_v1_enabled

---------

Co-authored-by: Zach Mueller <muellerzr@gmail.com>

* Add representation for Conv1D, for better output info.

* code format for Conv1D

* We add a __repr__ func for Conv1D, this allows the print (or output) of the model's info has a better description for Conv1D.

* enable

* fix