Glossary · Term

Llama-Factory

← all terms

Definition

Plain language

An open-source toolkit for fine-tuning Llama-family and other open language models.

As stated in the literature

A widely used open-source LLM fine-tuning library built on top of DeepSpeed and Hugging Face, supporting SFT, RLHF, and parameter-efficient methods.

Why it matters: It lowers the barrier to fine-tuning open models, which has been a major driver of the open-weight ecosystem's pace.

For example, a small team might use Llama-Factory to LoRA-tune Llama-3 on customer-support transcripts without writing their own training loop.

Heard on the show

“… And because DeepSpeed is the engine under three of the most popular SFT libraries — OpenRLHF, Llama-Factory, and TRL — every academic group running SFT through any of those libraries with CPU offloading …”
Episode 009 — How Two Silent Library Bugs Quietly Invalidated a Wave of Reasoning Papers

Mentioned in 1 episode

  1. 009
    How Two Silent Library Bugs Quietly Invalidated a Wave of Reasoning Papers

Related terms