List your company

LLM-Agent Oversight on ScienceWorld

0.245Scalar RegretScalar routing
0.232750.2388750.2450.251125Jun 19, 2026

Evaluation results

MethodLinks
Scalar routingIntervention Type=Scalar2026.060.245———
Action-conditioned (prefix) routingIntervention Type=Action-conditioned (prefix)2026.06—0.1690.0760.03