When experiencing a Hugging Face dataset download denial, the following options are available:
- Visit the Hugging Face account settings page to generate an Access Token
- Set environment variables in the terminal where OpenBench is running:
export HF_TOKEN="你的令牌" - For code space or container environments, it is recommended that the token be written to the
~/.bashrcor container configuration - Before rerunning the evaluation command, you can use the
huggingface-cli loginVerifying Token Validity - If evaluating a competition-level dataset such as GPQA, additional access to the dataset must be requested
It is recommended that token configuration be included in the project documentation to make it easier to handle permissions in a unified manner during team collaboration.
This answer comes from the articleOpenBench: an open source benchmarking tool for evaluating language modelsThe
May not be reproduced without permission:AI productivity tools " How do I resolve the issue of download permissions for benchmarking datasets?
English 
简体中文 
日本語 
Deutsch 
Português do Brasil