ResponseTooLong Error

[iankabujiya]

Occasionally, the library will return a longer response on the /search endpoint, which causes OpenAI to throw a ResponseTooLong error. Likely need to need to post-process JSON according to length restrictions of GPT function calling before returning to the calling model.

[iankabujiya]

Update: this is big challenging to get right w/o regression (since the API isn't always accurate on paging + shortening context degrades answer quality). Still looking at solutions.