Monitoring and Discovering Reward Hacking with Internal Representations during LLM Evaluations

Monitoring and Discovering Reward Hacking with Internal Representations during LLM Evaluations. It centres on Qwen, and also names Benchmarks and Reasoning Models. Reported by arXiv. Bharat Hunt files it under AI Models and Open Source AI — the section covering a new or updated model, its capabilities, benchmarks or availability.

Written by Bharat Hunt from the headline and the coverage below. The original reporting is the source of truth.