语义自一致性：通过语义加权增强语言模型推理
原标题：Semantic Self-Consistency: Enhancing Language Model Reasoning via Semantic Weighting
AI 摘要
该论文提出语义自一致性（Semantic Self-Consistency）方法，在传统自一致性框架基础上，对多条推理路径进行语义向量嵌入分析，而非仅对最终答案做多数投票。作者使用BERT系列模型（RoBERTa、SciBERT）对推理路径做向量化，通过语义加权和过滤机制重排结果、剔除退化或幻觉输出。在AQuA-RAT、SVAMP、StrategyQA三个数据集上，使用GPT-3.5、Llama 2/3、Mistral 7B、GPT-4o mini等模型验证，表明该方法提升了推理准确率和鲁棒性。
正文节选
Semantic Self-Consistency: Enhancing Language Model Reasoning via Semantic Weighting Abstract While large language models (LLMs) have rapidly improved their performance on a broad number of tasks, they still often fall short on reasoning tasks. As LLMs become more integrated in diverse real-world tasks, advancing their reasoning capabilities is crucial to their effectiveness in nuanced, complex problems. Wang et al. [33]’s self-consistency framework reveals that sampling multiple rationales befo