Petals - Easy way to run 100B+ language models at home

Run 100B+ language models at home, BitTorrent-style. Fine-tuning and inference up to 10x faster than offloading