Training Agents to Fail Predictably: Fine-Tuning for Forecastable Tail Risk

arXiv:2605.15134 · cs.LG · Submitted 2026-05-14 · Read on arXiv

cs.LG

Submitted: 2026-05-14

Updated: 2026-11-09

Code: https://github.com/unitaryai/detoxify

Terminology

Sources

Related papers