add test marker to run all tests with @require_bitsandbytes

[Titus-von-Koeller]

What does this PR do?

add test marker to run all tests with @require_bitsandbytes

Before submitting

• This PR fixes a typo or improves the docs (you can dismiss the other checks if that's the case).
• Did you read the contributor guideline,
  Pull Request section?
• Was this discussed/approved via a Github issue or the forum? Please add a link
  to it if that's the case.
• Did you make sure to update the documentation with your changes? Here are the
  documentation guidelines, and
  here are tips on formatting docstrings.
• Did you write any new necessary tests? Not needed IMO.

Who can review?

@younesbelkada, do I need any documentation for this and if yes, where?

[Titus-von-Koeller]

Submitting this as a separate PR, to keep things clean/atomic.

[younesbelkada]

Awesome, works for me, thanks a lot @Titus-von-Koeller !

[younesbelkada]

Thanks again ! Note to effectively run bnb tests it is sufficient to just run RUN_SLOW=1 pytest tests/quantization/bnb as running pytest -m bitsandbytes tests/ will run other tests such as tests that we put on llama etc which are already covered in tests/quantization/bnb, but this is a good addition anyway so I'd be keen to merge the PR

[ArthurZucker]

Thanks! let's just keep the same format for the markers

[HuggingFaceDocBuilderDev]

The docs for this PR live here. All of your documentation changes will be reflected on that endpoint. The docs are available until 30 days after the last update.

[Titus-von-Koeller]

Really strange, I just committed the marker description string change that @ArthurZucker suggested, but the tests are failing with unrelated stuff. Maybe the GH Runner doesn't have the right hardware for those tests?

FAILED examples/pytorch/test_pytorch_examples.py::ExamplesTests::test_generation - RuntimeError: "LayerNormKernelImpl" not implemented for 'Half'
FAILED examples/pytorch/test_pytorch_examples.py::ExamplesTests::test_run_audio_classification - ValueError: FP16 Mixed precision training with AMP or APEX (`--fp16`) and FP16 half precision evaluation (`--fp16_full_eval`) can only be used on CUDA or NPU devices or certain XPU devices (with IPEX).
FAILED examples/pytorch/test_pytorch_examples.py::ExamplesTests::test_run_glue - ValueError: FP16 Mixed precision training with AMP or APEX (`--fp16`) and FP16 half precision evaluation (`--fp16_full_eval`) can only be used on CUDA or NPU devices or certain XPU devices (with IPEX).
FAILED examples/pytorch/test_pytorch_examples.py::ExamplesTests::test_run_image_classification - ValueError: FP16 Mixed precision training with AMP or APEX (`--fp16`) and FP16 half precision evaluation (`--fp16_full_eval`) can only be used on CUDA or NPU devices or certain XPU devices (with IPEX).
FAILED examples/pytorch/test_pytorch_examples.py::ExamplesTests::test_run_semantic_segmentation - ValueError: FP16 Mixed precision training with AMP or APEX (`--fp16`) and FP16 half precision evaluation (`--fp16_full_eval`) can only be used on CUDA or NPU devices or certain XPU devices (with IPEX).
FAILED examples/pytorch/test_pytorch_examples.py::ExamplesTests::test_run_speech_recognition_ctc - ValueError: FP16 Mixed precision training with AMP or APEX (`--fp16`) and FP16 half precision evaluation (`--fp16_full_eval`) can only be used on CUDA or NPU devices or certain XPU devices (with IPEX).
FAILED examples/pytorch/test_pytorch_examples.py::ExamplesTests::test_run_speech_recognition_ctc_adapter - ValueError: FP16 Mixed precision training with AMP or APEX (`--fp16`) and FP16 half precision evaluation (`--fp16_full_eval`) can only be used on CUDA or NPU devices or certain XPU devices (with IPEX).
FAILED examples/pytorch/test_pytorch_examples.py::ExamplesTests::test_run_speech_recognition_seq2seq - ValueError: FP16 Mixed precision training with AMP or APEX (`--fp16`) and FP16 half precision evaluation (`--fp16_full_eval`) can only be used on CUDA or NPU devices or certain XPU devices (with IPEX).
FAILED examples/pytorch/test_pytorch_examples.py::ExamplesTests::test_run_vit_mae_pretraining - ValueError: FP16 Mixed precision training with AMP or APEX (`--fp16`) and FP16 half precision evaluation (`--fp16_full_eval`) can only be used on CUDA or NPU devices or certain XPU devices (with IPEX).

[ArthurZucker]

Yes it's unrelated to your PR don't worry!

[ArthurZucker]

rebasing on main should suffice to fix this !

[younesbelkada]

cc @ArthurZucker @ydshieh the CI runner is still failing - I think it uses the former cache that has the un-compatible naten library installed - #28432 IMO we can move forward and merge

[younesbelkada]

Hmm it seems to fail on other PRs too, e.g. :#28711

[ydshieh]

don't have good idea. i just re-triggered that job and let's see

[younesbelkada]

Thanks again !