LlamaFactory
by hiyouga
LlamaFactory is a toolkit for fine-tuning more than 100 open large language models and vision-language models, using either a command line or a web interface. It lets you adapt an existing model to your own data without writing training code.
- Stars
- 75.7k +2 a day
- Licence
- Apache-2.0
- Runs on
- A gpu machine, with a GPU
- Setup
- Technical
- Language
- Python
- Age
- 1248 days
What you can do with it
- Fine-tune open models such as Llama, Qwen, Gemma and DeepSeek on your own data using LoRA or QLoRA
- Train through the LLaMA Board web interface or a command-line config instead of writing code
- Serve the tuned model through an OpenAI-style API using vLLM or SGLang
Make money with it
A freelancer could offer custom model fine-tuning as a paid service, training and delivering a tailored model for clients' own data.
An idea, not financial advice. Check the licence before charging for anything built on it.
Ways a business could use it
- Train a small model on your own support tickets or product documents to answer customer questions without paying per-call fees to a hosted model
- Build a specialised model for a niche task, such as extracting data from invoices or documents, and run it on your own hardware
Can you use it commercially?
Free to use, change and sell, including in paid products. Keep the copyright notice.
Not legal advice. Read the project's licence file before building a product on it.
Getting it running
It is installed from source or with pip, or run through the provided Docker image, and needs a suitable Python and GPU environment; free Colab notebooks are also offered.
Follow the official instructions in the project README.
Things to check
- Needs a capable GPU or paid cloud GPU time for most models
- Documentation is described as work in progress
- Requires some machine-learning knowledge to prepare data and choose settings
Checked 27 Oct 2026. Summary drafted by AI from the project's README.
