"Dive into LLMs" is a series of programming practice tutorials for beginners in large models.
Purpose: This tutorial aims to provide an introductory programming reference related to large models, helping learners quickly get started with large models through simple practice and better carry out course design or academic research. The tutorial is expanded from the course handouts of "Frontier Technologies in Natural Language Processing" (NIS8021) and "Artificial Intelligence Security Technology" (NIS3353) at Shanghai Jiao Tong University, taught by Zhang Zhuosheng, and is public-interest in nature and completely free.
Main capabilities and content:
- Series of programming practice tutorials: Cover large model related topics in the form of programming practice. In the update on June 6, 2025, new topics such as mathematical reasoning, GUI Agent, large model alignment, and steganography were added, and the content of the original tutorial series was updated.
- Domestic public-interest tutorial "Full Process of Large Model Development": Launched by the team of this tutorial together with Huawei Ascend, based on Ascend basic software and hardware development, covering tutorial forms such as PPTs, lab manuals, and videos. This tutorial is divided into beginner, intermediate, and advanced series, oriented to different large model practice needs, and aims to provide, through code practice, a full-process development guide from easy to difficult for quickly getting started, applying models already supported by Ascend, and migrating and tuning entirely new models. You can visit the "Large Model Development Learning Zone" in the Ascend community to explore this series of courses.
How to use: The tutorial is organized as a GitHub repository, containing folders and files. Users can obtain the tutorial content through the repository and carry out programming practice learning. The project welcomes submitting Issues or PRs to participate in discussion.
Applicable scenarios: Suitable for students who wish to get started with large models, developers carrying out course design or academic research, and researchers and developers who need to practice large model development and model migration and tuning based on Ascend software and hardware.
Limitations and notes: This tutorial is an ongoing project, and omissions are unavoidable. All content comes only from contributors' personal experience, internet data, and related accumulation in daily scientific research work. All techniques are for reference only and are not guaranteed to be 100% correct. The badges used in this project come from the internet; if image copyright is infringed, please contact us for removal. Specific installation commands, code details, and other such materials are not provided in this material.
Contributors: "Dive into LLMs" series tutorial development team - Shanghai Jiao Tong University: Zhang Zhuosheng, Yuan Tongxin, Ma Xinbei, He Zhiwei, Du Wei, Zhao Haodong, Wu Zongru, Wu Zheng, Dong Lingzhong, Zhang Yulong. "Full Process of Large Model Development" series tutorial development team - Shanghai Jiao Tong University: Zhang Zhuosheng, Liu Gongshen, Chen Xingyu, Cheng Pengzhou, Dong Lingzhong, He Zhiwei, Ju Tianjie, Ma Xinbei, Wu Zheng, Wu Zongru, Yan Zihe, Yao Yao, Yuan Tongxin, Zhao Haodong; Huawei Ascend Community: ZOMI, Xie Qian, Cheng Liming, Lou Lihua,Jiao Zeyu.
