sp26-cs639 cs639
Lecture 22: Evaluation II Lecture 22: Evaluation II
Content locked
The catalog is public. Reading this content requires approved access.
Sign in to readLecture 22: Evaluation II
Beyond a single win rate
Rubric-based evaluation
Evaluate agents in an environment
Browser tasks: WebArena
Computer, code, and long-running tasks
Evaluate specific skills to locate failure modes
Needle in a haystack
Text-to-optimization
Connections
Source figure