小安

默认最大token是 4096，如果更高的话可能需要 GPT4，不是很建议，比较贵 > 将diff进行拆分提交来review 拆分会丢失一些上下文，可能效果不太好

为了节约token，只会对更新的文件进行review，如果文件改动过大（超过 4000 个字符）会自动跳过 review 看了下，OPENAI 确实改成了 3 次请求 / 分钟，之前我记得更多点有好几十，这里没做什么优化

@headless-kjh Actually I have no idea what causes this. could you help to re-run the job, let's check if it doesn't exit for the PR every time

会的，因为要掉用 Open AI 的接口，所以都会传到 Open AI 那边, 但 Open AI 有说不会使用用户数据用于训练模型 https://help.openai.com/en/articles/7039943-data-usage-for-consumer-services-faq > Please note that for our API product, OpenAI will not use data submitted by customers via our...

@HosakaKeigo sorry for the test is not workable yet. Could you help to create a pull request of beyond code, it looks fine to me.

@jeonbyeongmin Yes. This a good idea. Considering removed files are not Being used anymore, and review the removed/renamed files is no help on the code, and it's good for reducing...

@jeonbyeongmin Appreciated Bad new is the it's not easy to test. There are two ways 1. Using GitHub app, the code base is based on probot, you can search and...

@jeonbyeongmin i am guessing don't need to punish an action In my case, I am using anc95/ChatGPT-CodeReview@dev To test. It can be found and triggered You can try to replace...

@jeonbyeongmin Sorry for being so late to reply you I searched the document. there is no clear introduction for `files`, [here](https://docs.github.com/en/rest/commits/commits?apiVersion=2022-11-28#compare-two-commits) is the document for compare two commits. it says...

Thanks, @tgnemecek, I am planning to add some customizable functions too. for example custom prompt, custom model, and multiple language support, It's good for us to start this from a...