编译训练法：将自然语言规范转化为本地神经函数

AI 摘要

Hugging Face 论文页面介绍了一种名为“Compile by Training”的方法，通过将自然语言规范编译为可复用的本地神经函数，利用教师模型生成示例并微调小型适配器，避免了远程模型依赖。在 FuzzyBench-Hard 子集上，该方法达到 83.6% 的语义准确率，但编译时间约为一分钟，高于快速编译器。论文还展示了多站点网站助手、3D 头像控制和英-克劳德风格翻译等应用，并提供了公开交互服务。社区评论指出该方法可跳过每次规范微调，但存在教师示例覆盖不足和版本漂移的风险。

正文节选

Compile by Training: Turning Natural-Language Specifications into Local Neural Functions Abstract Compile by training converts natural-language specifications into reusable neural functions by distilling teacher-generated examples into small adapters, enabling efficient deployment without remote model dependencies. Many recurring text functions are easy to describe but difficult to implement with rules, while calling a large remote model for every input introduces repeated cost, latency, and dep

发布时间：—

抓取时间：2026-09-05 00:39

来源机构：Hugging Face