Evaluation short description

Aside from Multi-IF, there are very few multi-turn instruction following evals. MultiChallenge is a hard version that OpenAI and others report in their model cards.

Evaluation metadata

Provide all available

• Paper url: https://arxiv.org/abs/2501.17399
• Github url: https://github.com/ekwinox117/multi-challenge
• Dataset url: https://github.com/ekwinox117/multi-challenge/tree/main/data