Axolotl
Config-driven LLM fine-tuning for many models and techniques.
- Star 数
- 12.6K
- Fork 数
- 1.5K
- 未关闭 Issue
- 252
- 最近提交
- 2026-10-15
数据来自 GitHub API · 更新于 2026-10-15
Axolotl 是什么？
Axolotl streamlines fine-tuning with YAML configs covering full fine-tunes, LoRA/QLoRA, DPO and more, with multi-GPU support via DeepSpeed and FSDP.
主要功能
- YAML-configured training
- LoRA, QLoRA, full fine-tune, preference tuning
- DeepSpeed and FSDP
- Many dataset formats
系统要求
- 平台
- Linux · Docker
- 显卡
- NVIDIA (Ampere+ recommended)
- Python
- 3.11
- 备注
- PyTorch 2.5.1+.
如何安装 Axolotl
根据项目 README 整理。开源项目更新很快，请以 GitHub 页面上的最新说明为准。
pip
- 1Install with extras.
pip3 install --no-build-isolation axolotl[flash-attn,deepspeed] - 2Fetch examples and train.
axolotl fetch examples axolotl train examples/llama-3/lora-1b.yml
优势
- Reproducible configs
- Scales to multi-GPU
不足
- Linux-focused
Axolotl 替代品对比
| 软件 | Star 数 | 许可证 | 平台 | 显存 |
|---|---|---|---|---|
| Axolotl Config-driven LLM fine-tuning for many models and techniques. | 12.6K | Apache-2.0 | — | |
| Unsloth Fine-tune LLMs 2× faster with up to 70% less memory. | 78K | Apache-2.0 | 3 GB+ depending on model size | |
| LLaMA-Factory Unified fine-tuning for 100+ LLMs and VLMs with a no-code web UI. | 75.5K | Apache-2.0 | — | |
| TRL Hugging Face’s library for post-training LLMs: SFT, DPO, GRPO. | 19.5K | Apache-2.0 | — | |
| Kohya’s GUI GUI for training LoRA, DreamBooth and fine-tunes for SD, SDXL and FLUX. | 12.6K | Apache-2.0 | 8 GB+ for SD1.5 LoRA; 12–24 GB for SDXL/FLUX | |
| OneTrainer One-stop trainer for diffusion models with a desktop UI. | 3.2K | AGPL-3.0 | 8 GB+ |
常见问题
Axolotl 免费吗？
是的——Axolotl 是开源软件，采用 Apache-2.0 许可证。商用前请确认软件及模型权重的许可条款。
Axolotl 支持 Windows、Mac 和 Linux 吗？
Axolotl 支持：Linux, Docker。
运行 Axolotl 需要什么配置？
NVIDIA (Ampere+ recommended)
如何安装 Axolotl？
1. Install with extras. 2. Fetch examples and train.
Axolotl 有哪些好的替代品？
常见替代品包括 Unsloth, LLaMA-Factory, TRL, Kohya’s GUI，详见本页对比表。