liguodongiot/llm-action

Project overview

Shares technical principles and practical experience for large-language-model engineering through tutorials covering full fine-tuning, parameter-efficient fine-tuning, and RLHF. Several tutorials include accompanying code links, providing learning materials for developers and researchers studying LLM training.

Project type

Model Development

Use cases

Learning & Education

Deployment

Refer to project documentation

License

Apache-2.0

Best for

• Developers and researchers who need structured learning materials and practical examples for LLM training, full fine-tuning, and parameter-efficient fine-tuning.

Key capabilities

• Provides a curated set of practical LLM training tutorials spanning full fine-tuning, parameter-efficient fine-tuning, and RLHF.
• Provides practical guides for parameter-efficient fine-tuning techniques supported by HuggingFace PEFT, with accompanying code for several tutorials.

Limitations and risks

• The distributed training network communication section is marked as awaiting updates.
• GPU needs, minimum hardware, data boundaries, coding requirements, cost dependencies, and telemetry policies are not documented.

Getting started

• Setup difficulty, setup requirements, and first success paths are not documented.

Evidence and sources

• GitHub project description: 本项目旨在分享大模型相关技术原理以及实战经验（大模型工程化、大模型应用落地）
• README: 下面汇总了我在大模型实践中训练相关的所有教程。从6B到65B，从全量微调到高效微调（LoRA，QLoRA，P-Tuning v2），再到RLHF（基于人工反馈的强化学习）。
• README: 下面给大家分享**大模型参数高效微调技术实战**，该系列主要针对 HuggingFace PEFT 框架支持的一些高效微调技术进行讲解。
• README: 待更新...
• README: - Zero-Shot Prompting - Few-Shot Prompting - Chain-of-Thought (CoT) Prompting - Automatic Chain-of-Thought (Auto-CoT) Prompting - Tree-of-Thoughts (ToT) Prompting

Reviewed at: 2026-07-17T10:50:22Z