Open Models Almanac
Methodspost-trainingreinforcement learning algorithm

single source · post-training

PivotRL

A named reinforcement-learning method used on single-step agentic environments and in a distillation stage.

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.

Also called Pivot Reinforcement Learning.

Filed under post-training › reinforcement learning algorithm