[niutech]

Or maybe using Distributed Llama? https://github.com/b4rtaz/distributed-llama