AI agents now eval without labeled data
A new framework lets AI agents be evaluated without labeled benchmarks, using a stronger model as a proxy judge.
A new framework lets AI agents be evaluated without labeled benchmarks, using a stronger model as a proxy judge.