However from his examples (and his own admission) it seems that AutoGen isn't benefitting from full GPT4-level performance even tho he's pointed it directly at OpenAPI GPT4 (and other LLMs). The back and forth between the agents does not produce great results even tho similar prompts pumped directly into ChatGPT seem to give better results.
Comments
Matthew Berman has a good series on AutoGen with tutorials and demos: https://www.youtube.com/watch?v=10FCv-gCKug
However from his examples (and his own admission) it seems that AutoGen isn't benefitting from full GPT4-level performance even tho he's pointed it directly at OpenAPI GPT4 (and other LLMs). The back and forth between the agents does not produce great results even tho similar prompts pumped directly into ChatGPT seem to give better results.
Anyone know whats going on?
Temperature being set differently is one culprit. There are a few hyper parameters that can be tweaked to get some pretty different output.
This is a top potential cause for sure. The variability can change drastically with temperature differences