VRAM any HF model needs + live GPU rental prices
What a LoRA, QLoRA or full fine-tune costs, live prices