Early-Signal Detector

method · first seen 19 Oct 2026

co-evolving policy and training environment

Adapting the training task distribution, rollouts, and rewards jointly with the evolving policy.

Mentions per week

0119 May11 Aug10 NovPapersCodeDiscussion

PapersCodeDiscussion

Earliest sources

Latest sources

Mentioned together with

rubric-based reward