一作ACL 2026Poster· Natural Language Processing
Dissecting Failure Dynamics in Large Language Model Reasoning
Wei Zhu · Jian Zhang · Lixing Yu · Kun Yue · Zhiwen Tang
摘要
Large Language Models (LLMs) achieve strong performance through extended inference-time deliberation, yet how their reasoning failures arise remains poorly understood. By analyzing model-generated reasoning trajectories, we find that errors are not uniformly distributed but often originate from a...