Open Models Almanac
Methodspost-trainingreward modelling

single mention · post-training

Hallucination mitigation through abstention and context-sensitive RL

Uses abstention data and Merlin–Arthur-style RL environments to address hallucination.

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

Also called Hallucinations.

Filed under post-training › reward modelling