Megadose AI progress, ranked and analyzed.

When Models Edit Too Much: On the Fidelity of Minimal Code Edits

· HF Daily Papers ·
Strong code models can pass tests while making repairs larger and harder to review than necessary.

The paper builds repair tasks by adding controlled AST-level bugs to 400 BigCodeBench solutions, so each task has a known minimal patch. It finds over-editing across frontier LLMs, including GPT-5.5, with unnecessary text changes and added cognitive complexity. A preservation instruction reduced excess Levenshtein distance, cut added cognitive complexity by 26.6%, and slightly improved Pass@1. Reinforcement learning handled out-of-domain edit fidelity better than supervised fine-tuning. HF Daily Papers' note

score 5

Categories: Research