In this meeting between the State Changers, the team was looking at some issues they were experiencing with making multiple API calls to OpenAI. They discussed how their function was running a significant number of times within a short timeframe and speculated that this might be causing rate limiting issues.
They moved forward by examining data inputs and how data was exported, with most of it working smoothly with the exception of 'ad stack' which was running fewer times than expected. Further investigation revealed that they were not capturing all variants of the data due to the nature of incoming data from a procedural statistical system with inherent variation. One suggestion made was to log each API call to better track delays and errors, thus helping to gain more insight into when issues occur and how the system responds. This would allow them to compare requests and responses and make more informed decisions on error handling. Another possible approach discussed was to wrap the API call in a function that would handle error corrections, introducing a cooldown period when the system signals that it's been called too frequently before trying again. Lastly, they discussed the potential usefulness of making certain function calls within the XANO interface more visually distinctive for easier tracking. This would help with locating specific functions amidst a list of many types. Overall, the meeting was very technical and focused heavily on problem-solving for API call issues with OpenAI. The tools and platforms mentioned included OpenAI, XANO, API, and regex.
(Source: Office Hours 12/20 pm )
Join State Change Risk-Free