Distributed training of LLMs on consumer GPUs (via runxiyu) — discussion

#ai