Pass@k

Concept

A metric for evaluating LLMs, particularly in coding and mathematical problems, representing the probability that at least one of 'k' generated attempts is successful.

Mentioned in 1 video