search

IS2Lab / S-Eval

S-Eval: Automatic and Adaptive Test Generation for Benchmarking Safety Evaluation of Large Language Models
Other
42 stars 3 forks source link

Excellent work #1

Closed xiaoxunlong closed 5 months ago

xiaoxunlong commented 5 months ago

Excellent work! It is only 1k data in the dataset. When will the author release the total dataset?

IS2Lab commented 5 months ago

Thanks. We just released another 20000 attack prompts.

xiaoxunlong commented 5 months ago

Thank you for your great work!

xiaoxunlong commented 5 months ago

By the way, when will the 200k left data be released?

IS2Lab commented 5 months ago

Thanks for your recognition of our work. We have further released 10,000 base risk prompts and 100,000 corresponding attack prompts (Version-0.1.1). If our work is helpful to you, we would greatly appreciate it if you could clone and star the repository.