Diagnosing Harmful Continuation in Answer-Correct Long-CoT Training Traces
📰 ArXiv cs.AI
arXiv:2605.29288v1 Announce Type: new Abstract: Long chain-of-thought (CoT) traces are widely used as supervision for reasoning-oriented LLM SFT, yet answer-correct traces can still lead to markedly different fine-tuning outcomes. We study post-conclusion continuation in answer-correct long-CoT data: a continuation where the answer appears sufficiently supported, but the trace continues with additional reasoning that remains in the supervised target. To test its training effect, we use a delete-
Full Article
Title: Diagnosing Harmful Continuation in Answer-Correct Long-CoT Training Traces
Abstract:
arXiv:2605.29288v1 Announce Type: new Abstract: Long chain-of-thought (CoT) traces are widely used as supervision for reasoning-oriented LLM SFT, yet answer-correct traces can still lead to markedly different fine-tuning outcomes. We study post-conclusion continuation in answer-correct long-CoT data: a continuation where the answer appears sufficiently supported, but the trace continues with additional reasoning that remains in the supervised target. To test its training effect, we use a delete-
Abstract:
arXiv:2605.29288v1 Announce Type: new Abstract: Long chain-of-thought (CoT) traces are widely used as supervision for reasoning-oriented LLM SFT, yet answer-correct traces can still lead to markedly different fine-tuning outcomes. We study post-conclusion continuation in answer-correct long-CoT data: a continuation where the answer appears sufficiently supported, but the trace continues with additional reasoning that remains in the supervised target. To test its training effect, we use a delete-
DeepCamp AI