Open Models Almanac
Methodspost-trainingreinforcement learning algorithm

single mention ยท post-training

Effort-conditioned subgroup reward centering

Mean-centers rewards among responses sharing the same prompt and effort level to compute group-relative advantages.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See reinforcement learning algorithm for the methods filed beside it.

Filed under post-training โ€บ reinforcement learning algorithm