3 ms·

What are your thoughts on fine tuning vs low rank adaptations?

• brucethemoose2 3y ago
  Llama LORAs are very customizable, and range from "almost full finetuning" to "barely taking more VRAM than GPTQ inference"