Four open source projects dominate LLM fine-tuning today. Unsloth, Axolotl, TRL, and LLaMA-Factory all wrap the same underlying PyTorch and Hugging Face stack. They ...