# QLoRA

Combines 4-bit quantization of the frozen base with LoRA adapters, which cuts memory enough to fine-tune large models on one GPU.

## What it is used for

Fine-tuning large models on limited hardware.