Do you feel distributed LLM is the future? 
GitHub
GitHub - bigscience-workshop/petals: 🌸 Run LLMs at home, BitTorrent-style. Fine-tuning and inference up to 10x faster than offloading
🌸 Run LLMs at home, BitTorrent-style. Fine-tuning and inference up to 10x faster than offloading - bigscience-workshop/petals