How to fine-tune massive models on consumer GPUs by updating only a tiny fraction of parameters
14 views