" I recommend a book about large models from theory to practice, co-authored by Zhang Qi, Gui Tao, Zheng Rui, and Wan Xuanjing. Follow the official account to get the electronic version. "
Table of Contents
1 Introduction
1.1 Basic concepts of large-scale language models
1.2 Development history of large-scale language models
1.3 Large-scale language model construction process
1.4 Content arrangement of this book
2 major language model basics
2.1 Transformer model
2.2 Generative pre-training language model GPT
2.3 Large language model structure
3 Language model training data
3.1 Data source
3.2 Data processing
3.3 Data impact analysis
3.4 Open source data collection
4 Distributed training
4.1 Overview of distributed training
4.2 Distributed training parallel strategy
4.3 Cluster architecture for distributed training
4.4 DeepSpeed practice
5 Supervised fine-tuning
5.1 Prompt learning and contextual learning
5.2 Efficient model fine-tuning
5.3 Model context window expansion
5.4 Instruction data construction
5.5 Deepspeed-Chat SFT practice
6 Reinforcement Learning
6.1 Reinforcement learning based on human feedback
6.2 Reward model
6.3 Proximal strategy optimization
6.4 MOSS-RLHF practice
7 major language model applications
7.1 Inference planning
7.2 Comprehensive application framework
7.3 Intelligent agent
7.4 Multimodal large model
7.5 Large language model inference optimization
8 major language model evaluations
8.1 Overview of model evaluation
8.2 Large language model evaluation system
8.3 Large language model evaluation methods
8.4 Large Language Model Evaluation Practice
To get the e-book: follow the official account, automatically reply to the "AI Learning Materials" network disk address, enter the "1. Must-read for getting started" directory, and download "Large-Scale Language Model - From Theory to Practice - Zhang Qi-LLM-TAP.pdf" .
Reading recommendations:
Thinking and practice of AI technology: AI large model, technical details, application development
What can AI large model LLM do to help enterprises?
Recommend FinGLM, an open source large model project for financial analysis
A long article of 10,000 words will give you a comprehensive interpretation of the visual model
Embrace the future and learn AI skills! Follow me and receive free AI learning resources.