Rudeg··on LoRA from scratch: implementation for LLM finetuningnice, looks very cool and useful! I'll definitely try it!