I recently joined a hackathon in Hangzhou, where we had three days to create a product prototype. Our project, Rabbit Multi-Agents, is inspired by the Jade Rabbit from the Chinese myth of Chang'e flying to the moon: If a rabbit lived on the moon, it would need a plant factory to grow its food.
We wanted to see if an automated workflow powered by large language models (LLMs) could run a plant factory. Here's what we focused on:
- Handling and analyzing data: Testing the LLM's math skills and tool usage.
- Creating planting strategies: Seeing if the LLM could pull the right info from its knowledge base and give solid advice.
- Controlling equipment: Checking if the LLM could generate specific instructions and target values for the equipment.
We set up a workflow with three different AIs for data analysis, strategy creation, and command execution to avoid overwhelming the LLM.
After the event, we decided to keep the project open-source to attract other developers and improve it further. This is our way of giving back to the open-source community and helping others use this technology in agriculture or other fields. If you're interested, reach out to me!
Check it out at https://github.com/ThomasXIONG151215/RABBIT-Multi-Agents.
We'd love to hear your thoughts and get your support!
by Yuanke_Thomas
1 Comment
I’ve had horrible experiences trying to use LLMs for more complex math. They look like they’ve got it but it’s BS coming out.