huggingface / lighteval

Lighteval is your all-in-one toolkit for evaluating LLMs across multiple backends
MIT License
852 stars 100 forks source link

[EVAL] Update IFEval dataset #266

Closed clefourrier closed 3 months ago

clefourrier commented 3 months ago

Evaluation short description

Let's now use this implementation from the authors (ported by @lewtun ): https://huggingface.co/datasets/google/IFEval

lewtun commented 3 months ago

I also updated the one in the Eleuther harness for consistency: https://github.com/EleutherAI/lm-evaluation-harness/pull/2218