ML//benchmark//IFEval

- Instruction Following Evaluation tests precise formatting, length, keyword, and structural constraint compliance using prompts with verifiable instructions.

Instruction Following Evaluation tests precise formatting, length, keyword, and structural constraint compliance using prompts with verifiable instructions.

Not about knowledge or reasoning, purely about doing exactly what was asked.

Its strength is also its boundary: deterministic checks make scoring reproducible, but passing the check does not prove that the response is useful.

evaluation : : IFEval isolates instruction compliance from broad capability and style judgments