3 ms·

I do think it's a pretty cool demo, but I have to say I didn't love that the PR claims Sweep did "manual testing" of the fix. Additionally, sweep reviews the PR

I do think it's a pretty cool demo, but I have to say I didn't love that the PR claims Sweep did "manual testing" of the fix. Additionally, sweep reviews the PR and claims that the function is correctly implemented. A sibling points out that there's actually no testing added or done, and there's also issues with the implementation itself. This appears to be a general issue with GPT4 based products, they are extremely self-confident in their language. Presumably this stems from the overall training to work well as a chatbot. It's very cool that it inferred the right place to make the change and the steps of finding relevant code, making a plan, then doing it are things I wish all my junior developers would do! This is certainly moving in the right direction.

kevinlu1248 3y ago
Yup, it's a bit frustrating since it's a problem with LLM, RLHF and fine-tuning for chat. In fact, we also added in the prompt to not say that it did testing. I find that in general it seems really difficult to tell a language model (especially 3.5) to not do something. The self-review generally catches stuff like this since we tell it that this code is written by an inexperienced developer, so that Sweep becomes more critical.