Releases: huggingface/dataset-viewer
Releases · huggingface/dataset-viewer
0.8.1
Features:
- add "features" (column types) to /rows response. See https://huggingface.co/docs/datasets/about_dataset_features.html.
0.8.0
0.6.1
Fixes:
- the dataset
allenai/c4
is blocklisted to avoid blocking the app (#17 (comment) + only one worker in the app)
0.6.0
Breaking:
- the format of the response of the endpoints /datasets, /configs, /splits and /rows has changed.
- the behavior has changed if config or split are missing in the endpoints /splits and /rows.
Features:
- in /splits: if config is missing, all the splits of all the configs of the dataset are returned
- in /rows: if config is missing, all the rows of all the splits of all the configs of the dataset are returned
- in /rows: if split is missing, all the rows of all the splits of the config are returned
Details: 0.5.0...0.6.0
0.5.0
Breaking:
- minimum version of Python is now 3.9.6
- fix the number of workers to 1
Features:
- add endpoints:
/cache
,/datasets
- cache all the responses for
/datasets
,/info
,/configs
,/splits
,/rows
- environment variables can be setup in a .env file
- rename environment variables:
HOSTNAME
toAPP_HOSTNAME
andPORT
toAPP_PORT
- add environment variables:
CACHE_SIZE_LIMIT
,CACHE_TTL_SECONDS
,DATASETS_ENABLE_PRIVATE
,HF_TOKEN
,LOG_LEVEL
- add two targets for development:
make coverage
andmake watch
- prepare support for private datasets, but it's currently disabled (hardcoded in memorize)
CI:
- check the types with mypy
- ignore
safety
alert about tensorboard 2.6.0 - setup code coverage with codecov and pytest-cov
Refactor:
- refactor the benchmark to use the API instead of accessing the functions directly
- use
logging
to manage the logs
Details: 0.4.6...0.5.0
0.4.6
Features:
- feat: 🎸 upgrade datasets to get pathlib fix
Fixes:
- fix: 🐛 return the adequate status code in case of error
CI:
- add unit tests to the CI
Detail: 0.4.5...0.4.6
0.4.5
0.4.2
0.4.1
0.4.0
The body of 4xx errors is now a JSON: https://datasets-preview.huggingface.tech/splits?dataset=glue&config=NOSUCHCONFIG returns
{
"status_code": 404,
"exception": "Status404Error",
"message": "The dataset config could not be found.",
"cause": "ValueError",
"cause_message": "BuilderConfig NOSUCHCONFIG not found. Available: ['cola', 'sst2', 'mrpc', 'qqp', 'stsb', 'mnli', 'mnli_mismatched', 'mnli_matched', 'qnli', 'rte', 'wnli', 'ax']"
}