Language Models can perform Single-Utterance Self-Correction of Perturbed Reasoning
This addresses the problem of unreliable reasoning in AI for researchers and practitioners, though it appears incremental as it builds on existing self-correction concepts.
The paper tackles the brittleness of large language models in mathematical reasoning by showing they can self-correct synthetic perturbations in their reasoning chains, with robust single-utterance corrections observed across models and datasets.
Large Language Models (LLMs) have demonstrated impressive mathematical reasoning capabilities, yet their performance remains brittle to minor variations in problem description and prompting strategy. Furthermore, reasoning is vulnerable to sampling-induced errors which autoregressive models must primarily address using self-correction via additionally-generated tokens. To better understand self-correction capabilities of recent models, we conduct experiments measuring models' ability to self-correct synthetic perturbations introduced into their Chain of Thought (CoT) reasoning. We observe robust single-utterance intrinsic self-correction behavior across a range of open-weight models and datasets, ranging from subtle, implicit corrections to explicit acknowledgments and corrections of errors. Our findings suggest that LLMs, including those not finetuned for long CoT, may possess stronger intrinsic self-correction capabilities than commonly shown in the literature. The presence of this ability suggests that recent "reasoning" model work involves amplification of traits already meaningfully present in models.