ExL3-QLoRA trains 2-8 bpw exLlamaV3 quantized models using plain PyTorch; adapters saved in standard PEFT format

EXL3-QLoRA I've built this over the last few months and I think it's stable enough to share now. A fork of ExLlamaV3, this is a native ExLlamaV3 training implementation with plain PyTorch, without transformers....