← Glossary Company
Unsloth
An open-source project that makes fine-tuning and running language models faster and small enough for one GPU.
Unsloth is a small team and an open-source library aimed at a practical problem: adapting a language model to your own data usually needs more memory than a normal machine has. Unsloth rewrites the slow parts of that process so the same job fits on a single consumer graphics card and finishes several times faster.
It is also one of the fastest projects to publish ready-to-run, compressed versions of new open models, often within hours of release. If you have ever downloaded a quantised model from Hugging Face to run locally, there is a good chance it came from Unsloth.
Mentioned in