IFEval

IFEval (Instruction-Following Evaluation) is a straightforward benchmark that evaluates Large Language Models' ability to follow verifiable instructions such as word count requirements, formatting constraints, and specific content guidelines. This benchmark focuses on measurable instruction adherence through objective criteria, testing models' capability to understand and comply with explicit directives and constraints.