huggingface / datasets

🤗 The largest hub of ready-to-use datasets for ML models with fast, easy-to-use and efficient data manipulation tools
https://huggingface.co/docs/datasets
Apache License 2.0
19.18k stars 2.67k forks source link

`datasets-cli test` should work with datasets without scripts #3162

Open sashavor opened 2 years ago

sashavor commented 2 years ago

It would be really useful to be able to run datasets-cli testfor datasets that don't have scripts attached to them (whether the datasets are private or not).

I wasn't able to run the script for a private test dataset that I had created on the hub (https://huggingface.co/datasets/huggingface/DataMeasurementsTest/tree/main) -- although @lhoestq came to save the day!

hmtkvs commented 2 years ago

It would be really useful to be able to run datasets-cli testfor datasets that don't have scripts attached to them (whether the datasets are private or not).

I wasn't able to run the script for a private test dataset that I had created on the hub (https://huggingface.co/datasets/huggingface/DataMeasurementsTest/tree/main) -- although @lhoestq came to save the day!

why don't you try to share that info with people, so you can also save some days.

lhoestq commented 2 years ago

Hi ! You can run the command if you download the repository

git clone https://huggingface.co/datasets/huggingface/DataMeasurementsTest

and run the command

datasets-cli test DataMeasurementsTest/DataMeasurementsTest.py

(though on my side it doesn't manage to download the data since the dataset is private ^^)

hmtkvs commented 2 years ago

Hi ! You can run the command if you download the repository

git clone https://huggingface.co/datasets/huggingface/DataMeasurementsTest

and run the command

datasets-cli test DataMeasurementsTest/DataMeasurementsTest.py

(though on my side it doesn't manage to download the data since the dataset is private ^^)

Hi! Thanks for the info. git cannot find the repository. Do you know if they have depreciated these tests and created a new one?

lhoestq commented 2 years ago

I think it's become private, but feel free to try with any other dataset like lhoestq/test for example at https://huggingface.co/datasets/lhoestq/test

hmtkvs commented 2 years ago

I think it's become private, but feel free to try with any other dataset like lhoestq/test for example at https://huggingface.co/datasets/lhoestq/test

your example repo and this page https://huggingface.co/docs/datasets/add_dataset.html helped me to solve.. thanks a lot