Configurations in yaml not working

[tsor13]

Dataset configurations cannot be created in YAML/README

Hello! I'm trying to follow the docs here in order to create structure in my dataset as added from here (#5331):

datasets/docs/source/repository_structure.mdx

Lines 110 to 118 in 8b8e6ee

	```yaml
	---
	configs:
	- config_name: main_data
	data_files: "main_data.csv"
	- config_name: additional_data
	data_files: "additional_data.csv"
	---
	```

I have the exact example in my config file for my data repo:

configs:
- config_name: main_data
  data_files: "main_data.csv"
- config_name: additional_data
  data_files: "additional_data.csv"

Yet, I'm unable to load different configurations:

from datasets import get_dataset_config_names
get_dataset_config_names('tsor13/test', use_auth_token=True)

returns a single split, ['tsor13--test']

Does anyone have any insights?

@polinaeterna thank you for adding this feature, it is super useful. Do you happen to have any ideas?

Steps to reproduce the bug

from datasets import get_dataset_config_names
get_dataset_config_names('tsor13/test')

Expected behavior

I would expect there to be two splits, main_data and additional_data. However, only ['tsor13--test'] test is returned.

Environment info

• datasets version: 2.14.4
• Platform: macOS-13.4-arm64-arm-64bit
• Python version: 3.11.4
• Huggingface_hub version: 0.16.4
• PyArrow version: 12.0.1
• Pandas version: 1.5.1

[mariosasko]

Unfortunately, I cannot reproduce this behavior on my machine or Colab - the reproducer returns ['main_data', 'additional_data'] as expected.

[tsor13]

Thank you for looking into this, Mario. Is this on my repository, or on another one that you have reproduced? Would you mind pointing me to it if so?

[tsor13]

Whoa, in colab I received the correct behavior using my dataset. It must have something to do with my local copy of datasets (which again just failed).

I've tried uninstalling/reinstnalling to no avail

[polinaeterna]

hi @tsor13 , I haven't been able to reproduce your issue on tsor13/test dataset locally either. reinstalling doesn't help?