Early-Signal Detector

method · first seen 8 Oct 2026

KL regularization

Also: KL penalty, KL anchoring, KL-regularized RL, KL-regularized control, Kullback-Leibler regularization, KL-regularized reinforcement learning

Penalizing divergence from a reference policy during reinforcement learning fine-tuning of language models.

Mentions per week

0319 May11 Aug10 NovPapersCodeDiscussion

PapersCodeDiscussion

Rank 190 on 14 Nov 2026 · 6 mentions in 28 days · 5 independent groups · papers · first seen 5 weeks ago · acceleration +1.79

Earliest sources

Latest sources

Mentioned together with

offline reinforcement learninggroup relative policy optimizationdensity matrixoffline-to-online reinforcement learningsomatic hypermutationantibody affinity maturationsample complexityreinforcement learning from human feedbackpolicy gradientpessimistic value iteration