This project enables fast, memory-efficient LLM fine-tuning on Apple Silicon Macs without NVIDIA GPUs. LoRA adapters allow you to train and save only a few MB of parameters, making experimentation ...