English(EN) Generating Verifiable Chain of Thoughts from Exection-Traces

研究人员生成可验证的代码推理数据以提升LLM性能

作者 PulseAugur 编辑部 · [1 个来源] · 2026-04-28 04:00

研究人员开发了一种新方法，通过检测代码以捕获执行跟踪来生成可验证的代码推理思维链（CoT）解释。该流程将这些跟踪叙述成自然语言，并逐一将叙述与原始跟踪进行交叉检查以确保准确性。在经过验证的数据上微调模型，在代码推理和生成方面取得了显著的改进，在LiveCodeBench-Exec上的增幅高达+26.6。 AI

影响通过提供可验证的训练数据，改进了AI的代码推理和生成能力，有望带来更可靠的AI编码助手。

排序理由这是一篇详细介绍为AI模型生成可验证训练数据的新方法的学术论文。

在 arXiv cs.AI 阅读 →

AI 生成摘要 · Google Gemini · 来自 1 个来源。我们如何撰写摘要 →

本文如何被排名

Signal score: 0 / 100; Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket: Research; 这是一篇详细介绍为AI模型生成可验证训练数据的新方法的学术论文。
Source corroboration: Single-source cluster; Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics: paper, model release; Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance: High; Clearly on-topic for AI-industry coverage.
Story freshness: 126 days old; Aged out of breaking-news scoring windows; ranking reflects the durable signal from the full source set.

完整方法见我们的编辑标准。

报道来源 [1]

arXiv cs.AI TIER_1 English(EN) · Shailja Thakur, Vaibhav Saxena, Rohan Kulkarni, Shivdeep Singh, Parameswaran Selvam, Hima Patel, Hiroshi Kanayama · 2026-04-28 04:00

从执行跟踪生成可验证的思维链

arXiv:2512.00127v3 Announce Type: replace-cross Abstract: Getting language models to reason correctly about code requires training on data where each reasoning step can be checked. Current synthetic Chain-of-Thought (CoT) training data often consists of plausible-sounding explana…

本文如何被排名

报道来源 [1]

从执行跟踪生成可验证的思维链

相关实体

相关话题