The community member is trying to generate a "unified query framework" using the llamaindex tutorials, but is encountering a "Failed to select query engine" error during the testing phase. They are unsure whether they should try prompt engineering to choose one option or if there are other ways to resolve the issue. The comments suggest that using gpt-3.5 may be a factor, and that prompt engineering or trying gpt-4.0 could be potential solutions. However, there is no explicitly marked answer in the comments.
Hey guys, I am trying to generate a "unified query framework" (one of the llamaindex tutorials). During the testing phase, I am getting "Failed to select query engine" error (compare query). There are multiple choices being selected during the process. Should I be doing some sort of prompt engineering to just choose one? or is there any other ways that I could try? Thanks in advance.
For gpt-3.5, there might also be some prompt engineering involved. Tbh it's not that great at following instructions 😅 specifically the prompt for the selector I'm thinking 🤔
Yeah definitely, I've tried some prompt engineering to make it choose one option but chances of getting the right answer is like 50, 50. Maybe later I should try using gpt-4.0 😄 thanks for your quick response thou
