AI Research
Rethinking Generalization in Reasoning SFT: A Conditional Analysis on Optimization, Data, and Model Capability
A prevailing narrative in LLM post-training holds that supervised finetuning (SFT) memorizes while reinforcement learning (RL) generalizes. We revisit this claim for reasoning SFT with long chain-of-thought (CoT) supervision and find that cross-domain generalization is not absent...
AC
ArXiv CS.AI
Source
ArXiv CS.AI
AI Analysis
Generated insight by OnAirToday AI
This research challenges the notion that SFT inherently fails at generalization in LLMs, proposing instead a conditional view shaped by optimization, data, and model capability. Understanding these conditions is crucial for effectively improving LLM reasoning while managing potential trade-offs like safety.
Powered by AI • Not a substitute for human analysisMore Insights →
Share
Filed under:AI Research

Comments
Keep comments relevant and respectful. No links, advertising, self-promotion, or spam.