unsloth

Expert guidance for fast fine-tuning with Unsloth - 2–5× faster training, 50–80% less memory, LoRA/QLoRA optimization