LlamaFactory: Unified Efficient Fine-Tuning of 100+ Language Models

高效的微调对于将大型语言模型（LLM）适应到下游任务中至关重要。然而，要在不同的模型上实现这些方法需要付出相当大的努力。我们提出了LlamaFactory，这是一个统一的框架，集成了一套尖端的高效训练方法。它允许用户通过内置的Web UI LlamaBoard灵活地定制100多个LLM的微调，而无需编码。我们在语言建模和文本生成任务中经验证实了我们框架的效率和有效性。它已经在https://github.com/hiyouga/LLaMA-Factory发布，并已获得超过13,000个星和1,600个分支。

LlamaFactory: A Unified Framework for Efficient Fine-tuning of Large Language Models

The paper presents LlamaFactory, a unified framework that integrates a suite of efficient training methods for fine-tuning large language models (LLMs) without the need for coding.

The framework includes a built-in web UI called LlamaBoard that allows users to customize the fine-tuning of over 100 LLMs. The paper empirically validates the efficiency and effectiveness of the framework on language modeling and text generation tasks. The code has been released on GitHub and has received over 13,000 stars and 1,600 forks.

Related works include efficient fine-tuning methods for LLMs, such as distillation and pruning, as well as other frameworks for LLM training, such as Hugging Face's Transformers and Google's Lingvo.