RWKV used as model for intent classification followed by performing tasks on own (Similar to Auto-GPT).
Thanks for the efforts to develop this model. I have used this model and it performs very well especially the raven models.
As the title suggests, I wanted to know if the future RWKV models can be integrated with any exisiting application as an assistant to perform tasks on own based on the intents it predicts. It should output in a structured manner like a JSON like how AutoGPT does where it includes THOUGHTS, REASONING, CRITICISM and the ACTION to be performed.
P.S. - I am pretty excited about this research and want to know if any bigger model is scheduled to come out soon whose performance is comparable to commercial models out in the market.
I can train Raven with such data. Please tell everyone to collect GPT4/ChatGPT output data when using AutoGPT :)
So how would we share the data with you? And/or any plans to incorporate Openassistant training data?
Any plans to interface Raven with AgentLLM/AGIXT? https://github.com/Josh-XT/AGiXT/tree/main