研究人员开发了CANON，一种新颖的无标签自蒸馏方法，用于大型语言模型，该方法利用多个生成解决方案之间的共识来创建密集的、令牌级别的监督。这种方法在数学和科学基准测试上显著提高了推理准确性，将pass@1分数提高了多达12分。CANON以比现有的无标签强化学习方法少得多的计算量实现了这些提升，并证明了其解决模型以前无法解决的问题的能力。 AI
影响 在不需要标记数据的情况下增强LLM的推理能力，有可能降低训练成本并提高复杂任务的性能。
排序理由 该集群包含一篇详细介绍训练大型语言模型新方法的学术论文。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
arXiv:2607.13643v1 Announce Type: cross Abstract: Sampling multiple solutions and returning the majority answer is among the most reliable ways to improve the reasoning accuracy of large language models without labels, and a growing family of methods converts this consensus signa…
Sampling multiple solutions and returning the majority answer is among the most reliable ways to improve the reasoning accuracy of large language models without labels, and a growing family of methods converts this consensus signal into training supervision. However, existing app…
Sampling multiple solutions and returning the majority answer is among the most reliable ways to improve the reasoning accuracy of large language models without labels, and a growing family of methods converts this consensus signal into training supervision. However, existing app…