标题： 应用中国墙逆向工程技术进行大型语言模型代码编辑 显示英文标题

标题： Applying the Chinese Wall Reverse Engineering Technique to Large Language Model Code Editing

摘要： 大型语言模型用于代码（Code LLM）在编程环境中得到越来越多的应用。 尽管它们很有用，但顶级LLM的训练数据集仍未公开，这引发了关于潜在版权侵权的担忧。 一些模型，如Pleias和Comma，强调数据整理和许可证，然而，由于训练数据有限，这些模型并不具有竞争力，仅作为概念验证。 为了提高这些模型的实用性，我们提出应用“中国墙”技术，这一技术受到同名的逆向工程技术的启发——使用高质量模型为较弱模型生成详细指令。 通过这样做，可以使用一个较弱但符合伦理的模型来执行复杂任务，否则这些任务只能由更强大的模型完成。 在我们的评估中，我们发现该技术使Comma v0.1 1T在CanItEdit基准测试中的性能提高了超过66%，与单独运行同一模型相比，Starcoder2 Instruct的性能提高了大约20%。 然而，由于缺乏在无版权限制的公共领域内容上训练的模型，该技术目前的实际应用可能受到限制。 显示英文摘要

摘要： Large language models for code (Code LLM) are increasingly utilized in programming environments. Despite their utility, the training datasets for top LLM remain undisclosed, raising concerns about potential copyright violations. Some models, such as Pleias and Comma put emphasis on data curation and licenses, however, with limited training data these models are not competitive and only serve as proof of concepts. To improve the utility of these models, we propose an application of the "Chinese Wall" technique, inspired by the reverse engineering technique of the same name -- a high quality model is used to generate detailed instructions for a weaker model. By doing so, a weaker but ethically aligned model may be used to perform complicated tasks that, otherwise, can only be completed by more powerful models. In our evaluation, we've found that this technique improves Comma v0.1 1T's performance in CanItEdit benchmark by over 66%, and Starcoder2 Instruct by roughly 20% compared to when running the same model on the benchmark alone. The practical application of this technique today, however, may be limited due to the lack of models trained on public domain content without copyright restrictions.