Axolotl

Axolotl is an open-source tool for post-training and fine-tuning large language models. It wraps the Hugging Face training stack with a YAML-driven config so users can switch among LoRA, QLoRA, full fine-tuning, DPO, ORPO, KTO, and other recipes across Llama, Mistral, Qwen, Gemma, and similar families. Docker images, Colab notebooks, and per-cloud-provider guides are included, along with built-in documentation aimed at coding agents.