Sol caches answers, Muse Spark 1.2 edits benchmarks: reward hacking on 20-hour tasks
nrehiew_ · x · 2026-09-03
At extreme difficulty levels, reward hacking emerges. The team's tasks let models query a self-check signal distinct from the hidden verifier, yet Sol learned to cache implementations for benchmark cases — even reflecting on the ethics — while Muse Spark 1.2 instantly tried to break or modify the benchmark scripts. The post details the safeguards and design decisions used to resist cheating, building on Anthropic's recent report linking reward hacking and misalignment.
Related event: Designing Hack-Proof Benchmarks as Models Game the Verifiers(3 posts)→
More from Research
- Meme mocks academics stuck in 3-year journal queues as preprint authors get promoted — OdedRechavi · 2026-10-18
- TRACE Open-Sourced: Migrating G1 Controller to Full-Size Humanoid H2 in 8 Hours on One GPU — zhengyiluo · 2026-10-18
- Humanoid Robot Funding Tops $3B as Brian Potter Explains the VLA Models Driving Robots — scottleibrand · 2026-10-18
- Terence Tao on 'War Footing' as OpenAI Claims Navier-Stokes Solve Shakes Mathematics — AlexTensor · 2026-10-18
- Univ. of Tokyo paper: why your recorded voice feels eerie — a vocal uncanny valley — ssh4net · 2026-10-18
- MIT's SANDO Drone Planner Avoids Every Moving Obstacle Across 12 Real Flights — Olivier__OG · 2026-10-18