GitLab

* Properly guard PyTorch stuff

* [all-test]

* [all-test] Fix model imports as well

* Making sure StoppingCriteria is always defined

* [all-test]

* fix

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

* fix

* fix

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

* fix

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

* Add local agent

* Document LocalAgent

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

docs: ko: zero_shot_object_detection

Fix device issue in `SwiftFormerModelIntegrationTest::test_inference_image_classification_head` (#23435)

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

* fix

* fix

* fix

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

Update error

Fix + link

* Awesome Transformers

* Update

* Update

* Keywords

* Keywords

* Complete document

* Add lm-evaluation-harness

* Edit txtai according to David's comments

* Update awesome-transformers.md

* Create test_tokenization_electra.py

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

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

search model buffers for dtype as the last resort

Previously even after finding a stop token, other stop tokens were considered, which is unnecessary and slows down processing.

Currently, this unnecessary overhead is negligible since there are usually 2 stop tokens considered and they are fairly short, but in future it may become more expensive.

* fix sam slow test

* oops

* fix error message

* fix

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

It's more efficient to iterate over key, value dict pairs instead of iterating over keys and performing value lookups on each iteration. It's also more idiomatic.

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

* fix

* fix

* update

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

Logging an error and continuing is probably following the principle of least surprise.

It's more idiomatic and significantly more efficient because
1) it avoids repeated `append` call that Python has to resolve on each iteration
2) can preallocate the size of the final list avoiding resizing

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

* Add a test of the built release

* Polish everything

* Trigger CI

* [automodel] fix torch_dtype=auto in from_pretrained

* add test

* fix logic

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

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

* Fix translation

add conditional generation on docstring

* Few fixes in new Tools implementation

* code quality

* docs: ko: task/asr.mdx

* feat: manual draft

* fix: resolve suggestions
Co-authored-by: Hyeonseo Yun <0525yhs@gmail.com>

---------
Co-authored-by: Hyeonseo Yun <0525yhs@gmail.com>