JARVIS is a collaborative system for linking LLM and AI models.the systemLLM (Large Language Model) as controllerand many AI Models as Collaborative Executors(from HuggingFace Hub) Composition.
The workflow of the system consists of four stages:
- mission planning: Use ChatGPT to analyze user requests, understand their intent, and disassemble them into solvable tasks.
- model selection: To solve the planned task, ChatGPT chooses the AI model hosted on Hugging Face according to the description.
- task execution: Call and execute each selected model and return the result to ChatGPT.
- generate response: Finally, use ChatGPT to integrate the predictions of all models and generate a Response.
#JARVIS #Homepage #Documentation #Downloads #Collaborative #System #Connecting #LLM #Models #Development details