Library / SDK
Fine-tuning & Training
LLaMA-Factory
1Unified 2fine-tuning 3for 4hundred 5plus 6models
the six Ws · specification
W1
Who
Maintained by hiyouga and contributors.
W2
What
A unified framework for fine-tuning many LLMs and VLMs with LoRA, QLoRA, and preference optimization.
W3
Where
Runs self-hosted on GPUs via CLI and an optional web UI.
W4
When
Reach for it when you want a broad, beginner-friendly toolkit across many models.
W5
Why
Supports hundreds of models and techniques with both no-code and scriptable interfaces.
W6
With
GPUs (CUDA), PyTorch, model weights, and datasets.
W7
Watch
Community open-source (Apache-2.0); self-hostable and runs locally; GPU hardware required for training.