LLM Fine-tuning Framework
11.8k 2026-04-28
axolotl-ai-cloud/axolotl
A free and open-source framework designed for efficient fine-tuning of large language models.
Core Features
Supports a wide range of LLM architectures (e.g., Mistral, Qwen, GLM, Kimi).
Provides advanced fine-tuning techniques (e.g., LoRA, MoE expert quantization, GDPO, EAFT).
Optimized for VRAM reduction and distributed training (e.g., MoE expert quantization, Distributed Muon Optimizer).
Open-source and community-driven development.
Detailed Introduction
Axolotl is an open-source, comprehensive framework dedicated to simplifying and accelerating the fine-tuning process for large language models (LLMs). It provides robust support for various cutting-edge LLM architectures and incorporates advanced training methodologies like LoRA, MoE expert quantization, and distributed optimization. Designed for researchers and developers, Axolotl aims to make LLM customization more accessible and resource-efficient, enabling the creation of specialized models with reduced computational overhead.