QLoRA
LoRA on a base model stored in 4 bits (NF4), which fits much bigger models on small GPUs. It needs `bitsandbytes` and an NVIDIA GPU, and trains somewhat slower than plain LoRA.
LoRA on a base model stored in 4 bits (NF4), which fits much bigger models on small GPUs. It needs `bitsandbytes` and an NVIDIA GPU, and trains somewhat slower than plain LoRA.