Albert Villanova del Moral

Fix export to JSON when dataset larger than batch size

3

Fix export to JSON (`lines=False`) when dataset larger than batch size. Fix #7037.

Update datasets to 2.21.0

2

Update datasets to 2.21.0. Fix #3024.

Version 3 of upload-artifact GH Action is deprecated

2

The GitHub Action upload-artifact@v3 is deprecated and will raise an error on November 30, 2024: https://github.blog/changelog/2024-04-16-deprecation-notice-v3-of-the-artifact-actions/ > Starting November 30, 2024, GitHub Actions customers will no longer be able to...

Gradio Progress component does not work for HTML output

1

### Describe the bug Gradio Progress component does not work when the output of the function is HTML. ### Have you searched existing issues? 🔎 - [X] I have searched...

Requoting unquotes %27

5

### Describe the bug When requoting a redirection URL, you use `yarl.URL` and the single quotation mark gets unquoted: `%27` => `'` https://github.com/aio-libs/aiohttp/blob/4635161ee8e7ad321cca46e01ce5bfeb1ad8bf26/aiohttp/client.py#L578-L580 ```python In [1]: from yarl import URL...

CI parity test is broken due to FileNotFoundError: https://github.com/huggingface/evaluate/actions/runs/10827111652/job/30039577026?pr=631 ``` FAILED tests/test_trainer_evaluator_parity.py::TestEvaluatorTrainerParity::test_token_classification_parity - FileNotFoundError: [Errno 2] No such file or directory: '/tmp/tmpg8va97zxevaluator_trainer_parity_test/tokenclassification_conll2003_transformers/eval_results.json' ``` ``` _________ TestEvaluatorTrainerParity.test_token_classification_parity __________ [gw0] linux --...

Remove dead code `MockStreamingDownloadManager`. While working in a previous PR, I discovered this dead code.

Fix attribute and parameter names in loggers

1

Fix attribute and parameter names in loggers.

Use config dict in TaskConfigLogger for easier serialization

6

Use config dict (instead of dataclass) in TaskConfigLogger for easier serialization (e.g. json.dump). Note that dataclasses are not JSON serializable out of the box.

[FT] Fail faster when passing unsupported metrics to InferenceEndpointModel

2

## Issue encountered When passing an unsupported metric (e.g. single-token metric) to `InferenceEndpointModel`, an error is raised from `Pipeline.evaluate`: ``` ValueError: Endpoint models can't use single token metrics. Change the...