PreciseCoder: unit-test-only RL teaches code models to over-edit and hallucinate bugs
Berkeley · hf · 2026-10-16
Researchers from USC, UChicago, Google and UC Berkeley show that RL from unit tests raises pass rates but teaches models to edit more and hallucinate bugs in already-correct code. PreciseCoder augments the reward with semantically verified, localized edits: across 4B/9B/27B models it cuts average edited lines from 7–15 to 3 at comparable pass rates, and generalizes from single-line bugs to repository-level repair with smaller patches resolving more issues.
More from Research
- MathEconLib open-sourced: Axiom Math and top economists bring machine-checked proofs to economics in Lean 4 — soumitrashukla9 · 2026-10-23
- Ken Ono invites the community to build MathEcon on the new Lean 4 economics library — soumitrashukla9 · 2026-10-23
- IFM open-sources RL360: 8,192 concurrent sandboxes and ~1.3M rollouts for agentic RL — rupspace · 2026-10-23
- Science lifts expression of concern on 2024 AI debunking-conspiracies paper; conclusions unchanged — DG_Rand · 2026-10-23
- Embodied First Exam benchmarks coding agents on 1,090 tasks across 44 robots — ManlingLi_ · 2026-10-23
- Anima Anandkumar on neural operators: AI weather models beat supercomputers — AnimaAnandkumar · 2026-10-23