keventmesh / llm-tool-provider

Teach your LLM to dynamically call your services in Kubernetes
7 stars 3 forks source link

feat: improve should continue logic #16

Closed Cali0707 closed 4 months ago

Cali0707 commented 5 months ago

This PR aims to improve the "should continue" logic of the chat app. But, since this is so central to the whole thing it would be good to test it on some different prompts to make sure it all works still. With this new change, the model is able to do the following: image

Cali0707 commented 5 months ago

Let's hold of on merging this, it seems to struggle with questions when it does not know the answer now and just keeps cycling and prompting itself

Cali0707 commented 4 months ago

Superceded by #26