Axolotl

Axolotl is a free and open-source tool designed for efficient post-training and fine-tuning of large language models (LLMs). It supports a wide range of models like GPT-OSS, LLaMA, and Mistral, and offers multimodal training for vision-language and audio models. With easy configuration using a single YAML file, Axolotl simplifies the entire pipeline from dataset preprocessing to inference.

Key features include:

• Multiple Model Support: Train diverse models available on the Hugging Face Hub.
• Flexible Training Methods: Options like LoRA, QLoRA, and Reward Modelling.
• Cloud Ready: Compatible with cloud platforms and local hardware.
• Performance Optimizations: Includes advanced techniques for enhanced training efficiency.

Installation is straightforward with options for pip or Docker, and it requires an NVIDIA or AMD GPU. Axolotl is ideal for developers looking to fine-tune LLMs quickly and effectively.