With release 0.5.0, PEFT now officially supports fine tuning quantized GPTQ models! This is a pretty big deal as it allows you to download a much smaller model for fine tuning!