Self-Evaluating LLMs for Multi-Step Tasks: Stepwise Confidence Estimation for Failure Detection
This addresses the need for trustworthy LLM deployment in high-stakes, multi-step reasoning scenarios, though it is incremental as it builds on prior self-evaluation methods.
The paper tackled the problem of reliability and failure detection in large language models (LLMs) for multi-step reasoning tasks by extending self-evaluation techniques to such tasks, showing that stepwise evaluation outperforms holistic scoring with up to a 15% relative increase in AUC-ROC.
Reliability and failure detection of large language models (LLMs) is critical for their deployment in high-stakes, multi-step reasoning tasks. Prior work explores confidence estimation for self-evaluating LLM-scorer systems, with confidence scorers estimating the likelihood of errors in LLM responses. However, most methods focus on single-step outputs and overlook the challenges of multi-step reasoning. In this work, we extend self-evaluation techniques to multi-step tasks, testing two intuitive approaches: holistic scoring and step-by-step scoring. Using two multi-step benchmark datasets, we show that stepwise evaluation generally outperforms holistic scoring in detecting potential errors, with up to 15% relative increase in AUC-ROC. Our findings demonstrate that self-evaluating LLM systems provide meaningful confidence estimates in complex reasoning, improving their trustworthiness and providing a practical framework for failure detection.