AI RESEARCH
FABSVer: Faster Training and Better Self-Verification for LLM Mathematical Reasoning
arXiv CS.CL
•
ArXi:2605.28389v1 Announce Type: new While large language models have made significant progress in mathematical reasoning, they remain unreliable at judging the correctness of their own solutions. Existing approaches that equip models with self-verification typically treat solution generation and verification as two separate tasks, leading to substantially increased