Post #4256727

It may take a smaller model a lot longer to problem solve, though it is improving being augmented. I wanted to play with a larger model communicating with a smaller model. I am using granite 8 or 5 it can write code but it can't write things it doesn't know sadly. Then we add a simple rag layer to query GitHub stack overflow. What is weird is when the llm is told it won't work after awhile by a logic gate it made up functions to try and beat the logic gate that kept saying no lol