LLaMA-Factory is an open-source framework for fine-tuning and adapting large language and multimodal models. It provides a unified environment for training many model families with methods such as full fine-tuning, LoRA and other parameter-efficient techniques, preference optimization, quantization-aware workflows, and distributed training.

The project includes command-line tools and a web interface that simplify dataset configuration, training, evaluation, and export. By supporting a wide range of popular open models and training approaches in one toolkit, LLaMA-Factory is useful for researchers and developers who want to customize models for domain-specific tasks without assembling a separate training stack for every architecture.