作者位次 #2/6ACL 2026Findings· Findings of ACL
R eflexi C oder: Teaching Large Language Models to Self-Reflect on Generated Code and Self-Correct It via Reinforcement Learning
Juyong Jiang · Jiasi Shen · Sunghun Kim · Kang Min Yoo · Jeonghoon Kim · Sungju Kim
摘要
While Large Language Models (LLMs) have revolutionized code generation, standard “System 1” approaches that generate solutions in a single forward pass often hit a performance ceiling on complex algorithmic tasks. Existing iterative refinement strategies attempt to bridge this gap at inference ti...