Open Models Almanac
Methodspost-trainingrollout & RL infrastructure

single mention · post-training

Staleness-aware truncated importance sampling

Uses staleness-aware truncated importance sampling with partial rollout to accelerate training while preserving model quality.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See rollout & RL infrastructure for the methods filed beside it.

Filed under post-training › rollout & RL infrastructure