3 ms·
Created This Library for Finite State Machine Based LLM Agents
Creating this library where rather than defining single prompt and set of tools (like function based approach) of creating llm agents, instead you just define some set of states and individual prompts for each of them..
I basically made this to help the problems of hallucination and wrong function calls in my other project, but I urge every llm enthusiast to at least checkout once and give feedbacks,
here - https://github.com/searchX/moorellm
- gus_massa 2y agoClicky: https://github.com/searchX/moorellm https://github.com/searchX/moorellm
- hari01584 2y agothank you man
- eightysixfour 2y agoNeat OP - have you seen the autogen FSM implementation? How do you think it compares? https://microsoft.github.io/autogen/blog/2024/02/11/FSM-GroupChat https://microsoft.github.io/autogen/blog/2024/02/11/FSM-Grou...
- hari01584 2y agoI have seen autogen but I think it's driving principles are little different, auto-gen is more like a multi-agent approach.. Well we are also like that but moorellm driving principle is more like One agent, multiple states
- hari01584 2y agothough, pretty much sure autogen can do what we trying to do here, but again all those abstractions and terms are confusing...