Open Models Almanac
Methodsevaluationaveraging - mean@k; avg@k; pass@k; pass^k

single mention · evaluation

Pass@k

Computes pass@k across multiple sampled completions per prompt as a proxy for environment climb-ability.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See averaging - mean@k; avg@k; pass@k; pass^k for the methods filed beside it.

Also called pass@k plot.

Filed under evaluation › averaging - mean@k; avg@k; pass@k; pass^k