Symbol k

The passage around this formula

…is designed to punish an agent whose competence is real but inconsistent across resampled trials — a stochastic policy with true per-trial success probability p has $\text{pass}^k$ $\approx$ $p^k$ , which falls quickly as k grows. But a policy whose output on a given task is deterministic — an empty response, always, regardless of sampling — produces the identical transcript on every trial, so $\text{pass}^k_{\text{deterministic}} = \text{pass}^1 \quad \text{for every } k$. A degenerate exploit is not merely invisible to pass@1; it is more invisible to pas$s^k$, precisely…

Read this part in the article →