Radar

hiyouga/LlamaFactory

Unified Efficient Fine-Tuning of 100+ LLMs & VLMs (ACL 2024)

Python74.9k starsOtherView on GitHub

Star growth

153

stars/week · 28d

Total stars

74.9k

Other

Maintenance

Last commit
6d ago
2026-09-14
Last release
113d ago
2026-05-30
License
Apache-2.0

Read from GitHub when this page was built, 2026-09-20. Commit date is the head of the default branch, not the last push to any branch.

Why we are watching

Fine-tuning and adapting open-weight models on hardware you control, with the efficiency methods (LoRA/QLoRA, quantised training) that decide whether that is possible on one GPU or needs a datacentre. That is the efficiency-and-open-weights half of the theme: it produces a checkpoint you hold, rather than a fine-tune that lives inside a provider's account.

Measured as of August 6, 2026.