single mention · post-training
Fail-to-pass test verifier construction
Uses tests that fail before and pass after the reference solution as the task verifier.
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.
Filed under post-training › reward modelling
Explainer · D3 · 30 minSynthetic data: what models write for other modelsA generator writes what you can't crawl. It passes on its errors and biases, so what counts is whether anything can catch themExplainer · S4 · 25 minRLHF/RLVR: reward-based learning, from human preferences to verifiersA program that checks the answer is cheap to run and gives the same verdict every time, but exists only for checkable tasks; a learned judge reaches open-ended ones and gets exploited by the policy it trains