LLaMA-Factory alternatives
3 substitutes · 3 in training · 0 adjacent
The short answer
3 tools on this index are recorded as a substitute for LLaMA-Factory. 3 of these run the same way you already deploy LLaMA-Factory (self-hosted), and 3 are permissively licensed, so they do not change what you can ship. All of them sit in Fine-tuning & Training, so the choice is a real one rather than a category change. There is no overall winner here: which one is right depends on traffic shape, operational budget and how much of LLaMA-Factory you have already built around.
By Lattice editorial · Comparison reviewed · method
Real substitutes
in Fine-tuning & Training- Axolotllibrary · Apache-2.0
Configuration-driven fine-tuning across common architectures.
Use when Configuration-driven fine-tuning across many architectures.
Skip when You need to hand-tune the training loop itself.
- TRLlibrary · Apache-2.0
Hugging Face library of post-training trainers for SFT, DPO and GRPO.
Use when Post-training trainers for SFT, DPO and GRPO on top of any PEFT setup.
Skip when You are training from scratch rather than aligning an existing model.
- torchtunelibrary · BSD-3-Clause
PyTorch-native recipes for fine-tuning and aligning open models.
Use when PyTorch-native recipes that you can read and modify.
Skip when You want the breadth of a config-driven stack.
This graph is bidirectional and unsponsored. A tool appears here if LLaMA-Factory or another entry names it as a substitute, and Axolotl, TRL and torchtune point at LLaMA-Factory as well. How this index is compiled — including where it is wrong.