This course offers a comprehensive study of Natural Language Processing (NLP). We will delve into word representations, the construction of language models, the application of deep learning in NLP, and Large Language Models (LLMs). Specifically, we will discuss the architectural engineering, data engineering, prompt engineering, training techniques, and efficiency enhancements of LLMs. Students will gain insights into the application of NLP and LLMs in various domains, such as text mining, search engines, human-machine interaction, medical-legal consulting, low-resource languages, and AI for Science, as well as how to handle issues of data privacy, bias, and ethics. The course will also investigate the limitations of NLP and LLMs, such as challenges in alignment. The curriculum includes guest lectures on advanced topics and in-class presentations to stimulate practical understanding. This course is ideal for anyone seeking to master the use of NLP and LLMs in their field.
这门课程提供了对自然语言处理(NLP)的全面研究。我们将深入探讨词表示、语言模型的构建、深度学习在NLP中的应用和大型语言模型(LLMs)。特别地,我们将讨论大型语言模型的架构工程、数据工程、提示工程、训练技巧以及效率提升。学生将了解到自然语言处理和大型语言模型在各个领域的应用,如文本挖掘、搜索引擎、人机交互、医疗法律咨询、低资源语言和AI for Science,以及如何处理数据隐私、偏见和伦理问题。课程还将研究NLP和LLMs的局限性以及模型对齐问题。该课程包括高级主题的客座讲座和课堂演示。

Teaching team


Instructor
Benyou Wang

Benyou Wang is an assistant professor in the School of Data Science, The Chinese University of Hong Kong, Shenzhen. He has achieved several notable awards, including the Best Paper Nomination Award in SIGIR 2017, Best Explainable NLP Paper in NAACL 2019, Best Paper in NLPCC 2022, Marie Curie Fellowship, Huawei Spark Award. His primary focus is on large language models.

WU Minghao
吴铭昊 / WU Minghao
TA
225040492@link.cuhk.edu.cn
Office hour: Friday 2:00-3:00 PM
Tencent Meeting: 523-927-1424
Research: LLM Agent, multimodal generation, world models
CHEN Shunian
陈舒年 / CHEN Shunian
TA
117010022@link.cuhk.edu.cn
Office hour: Tuesday 17:00-18:00
Tencent Meeting: 579-4322-1432
Research: Multimodal Large Language Models, Embodied AI
JI Ke
冀轲 / JI Ke
TA
224040370@link.cuhk.edu.cn
Office hour: Thursday 16:00-17:00
Tencent Meeting: 403-9123-6065
Research: LLM Agents, Recursive Self-Improvement
LI Sirui
李思睿 / LI Sirui
TA
223040027@link.cuhk.edu.cn
Office hour: Wednesday 2:00-7:00 PM
In person: TXB122
Research: Spiking Neural Network
WANG Kuang
王匡 / WANG Kuang
TA
224040348@link.cuhk.edu.cn
Office hour: Monday 10:00-11:00
Tencent Meeting: 616-4036-2524
Research: Multimodal Large Language Models, Interactive AI
LIANG Juhao
梁钜豪 / LIANG Juhao
TA
223010147@link.cuhk.edu.cn
Office hour: Monday 16:00-17:00
Tencent Meeting: 481-4890-5355
Research: 模型推理优化,多语言模型
RAO Siqi
饶思齐 / RAO Siqi
TA
226045029@link.cuhk.edu.cn
Office hour: Thursday 11:00 AM-12:00 PM
In person: Daoyuan 224
Tencent Meeting: 575-0182-9386
Research: LLM Agents, Responsible AI, Human Value Alignment
QIN Aoxiang
秦奥翔 / QIN Aoxiang
TA
224040222@link.cuhk.edu.cn
Office hour: Friday 3:00-4:00 PM
In person: Daoyuan 224
Tencent Meeting: 882-5850-5999
Research: LLM and agent alignment, LLM reasoning, long-context modeling

Logistics


Academic Calendar


First Term7 September 2026 (Mon) - 18 December 2026 (Fri)
Add/Drop6 September 2026 (Sun) - 18 September 2026 (Fri)
Course examinations20 December 2026 (Sun) - 27 December 2026 (Sun)
Holiday notice: Mid-Autumn Festival holiday is 25-27 September. National Day holiday is 1-7 October. 20 September (Sun) and 10 October (Sat) are designated working days. Classes scheduled during the holiday period are postponed accordingly.

Schedule


Date Topics Recommended Reading Slides Video Coding Events Deadlines
Sep. 7-10 Warmup Tutorial 0: GitHub, LaTeX, Colab, and ChatGPT API OpenAI's blog
LaTeX and Overleaf
Colab
GitHub
[Slides]
Sep. 11th Lecture 1: Introduction to NLP Hugging Face NLP Course
LLM Course: roadmaps and Colab notebooks
On the Opportunities and Risks of Foundation Models
Sparks of Artificial General Intelligence: Early experiments with GPT-4
[Slides] Lecture 1 video lecture90 min [Phoenix]
Sep. 18th Lecture 2: Basics of Linguistics and Word Representation Universal Stanford Dependencies: A cross-linguistic typology
Insights between NLP and Linguistics
Efficient Estimation of Word Representations in Vector Space (word2vec)
GloVe: Global Vectors for Word Representation
Evaluation methods for unsupervised word embeddings
[Slides] Lecture 2 video lecture91 min [Linguistics repo], [gensim] Assignment 1 out
[Assignment page], [PDF], [Colab]
Sep. 18th Tutorial 1: Introduction to PyTorch and Word Embeddings PyTorch: Learn the Basics
Enriching Word Vectors with Subword Information (FastText)
[Tutorial page] [Notebook 1: PyTorch], [Notebook 2: Word embeddings]
Sep. 25th Mid-Autumn Festival Holiday — No Class Mid-Autumn Festival holiday: Sep. 25 (Fri) – Sep. 27 (Sun)
Oct. 2nd National Day Holiday — No Class National Day holiday: Oct. 1 (Thu) – Oct. 7 (Wed)
Oct. 9th Lecture 3: Language Modeling A Neural Probabilistic Language Model
BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
Language Models are Few-Shot Learners (GPT-3)
[Slides] Lecture 3 video lecture87 min [BERT], [nanoGPT]
Oct. 16th Lecture 4: Transformer and Beyond Attention Is All You Need
HuggingFace's course on Transformers
Scaling Laws for Neural Language Models
The Transformer Family Version 2.0
On Position Embeddings in BERT
[Slides] Lecture 4 video lecture137 min [Annotated Transformer]
Oct. 23rd Lecture 5: Large Language Models (LLMs) LLaMA: Open and Efficient Foundation Language Models
Llama 2: Open Foundation and Fine-Tuned Chat Models
Qwen3 Technical Report
DeepSeek-V3 Technical Report
[Slides] Lecture 5 video lecture154 min [Fine-tune Llama 2] Assignment 1 due (11:59pm)
Assignment 2 out
[Assignment page], [PDF], [Colab]
Oct. 23rd Tutorial 2: Training Your Own LLMs — SFT, LoRA/QLoRA, PPO, GRPO and DPO LoRA: Low-Rank Adaptation of Large Language Models
QLoRA: Efficient Finetuning of Quantized LLMs
Direct Preference Optimization (DPO)
HF Cookbook: Post-training an LLM for reasoning with GRPO in TRL
[Tutorial page], [Slides], [AI tools for research] [Notebook A: SFT + RM + PPO], [Notebook B: GRPO], [Notebook C: DPO], [LLMZoo], [nanoGPT], [LLMFactory]
Oct. 30th Lecture 6: RL for LLM Post-Training — From Alignment to Reasoning Training language models to follow instructions with human feedback (InstructGPT)
Proximal Policy Optimization Algorithms (PPO)
Direct Preference Optimization (DPO)
ReMax: A Simple, Effective, and Efficient RL Method for Aligning LLMs
DeepSeekMath (GRPO)
DeepSeek-R1: Incentivizing Reasoning Capability in LLMs via RL
DAPO: An Open-Source LLM RL System at Scale
SimpleRL-Zoo: Zero RL for Open Base Models
[Slides] Lecture 6 video lecture152 min [verl], [TRL], [OpenRLHF], [ReMax], [SimpleRL] Final Project out
[Project guide & tracks], [PDF], [LaTeX source]
Oct. 30th Tutorial 3: Prompt Engineering and LLM Agents [Colab]
Nov. 6th Lecture 7: LLM Agents and Harness Engineering LLM Powered Autonomous Agents (Lilian Weng)
ReAct: Synergizing Reasoning and Acting in Language Models
Building Effective Agents (Anthropic)
Effective Context Engineering for AI Agents (Anthropic)
Effective Harnesses for Long-Running Agents (Anthropic)
Harness Engineering: Leveraging Codex in an Agent-First World (OpenAI)
Harness Engineering for Coding Agent Users (Böckeler)
Model Context Protocol
Agent2Agent (A2A) Protocol
Stanford CS329Z
[Slides] Lecture 7 video lecture131 min [Codex CLI], [OpenClaw], [MCP servers], [LangGraph] Assignment 2 due (11:59pm)
Assignment 3 out
Nov. 13th Lecture 8: Final Projects and Research Sharing ToolBench
Cline
Roo Code
[Slides], [Project guide page (EN/中文)], [Writing tips & samples] Lecture 8 video lecture77 min
Nov. 13th Tutorial 4: RAG and Agent Development [Colab] Optional: Project Proposal due
Team Formation due
Nov. 20th Lecture 9: Medical LLMs — Towards Generalist Medical AI HuatuoGPT, towards Taming Language Model to Be a Doctor
HuatuoGPT-II, One-stage Training for Medical Adaption of LLMs
HuatuoGPT-o1, Towards Medical Complex Reasoning with LLMs
Large Language Models Encode Clinical Knowledge (Med-PaLM)
Towards Conversational Diagnostic AI (AMIE)
Agent Hospital: A Simulacrum of Hospital with Evolvable Medical Agents
Apollo: Multilingual Medical LLMs
ShizhenGPT: Multimodal LLMs for Traditional Chinese Medicine
[Slides] Lecture 9 video lecture161 min [HuatuoGPT], [HuatuoGPT-o1], [Apollo]
Nov. 27th Lecture 10: Multimodal LLMs Blog post: Generalized Visual Language Models
CLIP: Learning Transferable Visual Models From Natural Language Supervision
LLaVA: Visual Instruction Tuning
Moshi: a speech-text foundation model for real-time dialogue
Qwen2.5-Omni Technical Report
OpenVLA: An Open-Source Vision-Language-Action Model
π0: A Vision-Language-Action Flow Model for General Robot Control
VGGT: Visual Geometry Grounded Transformer
[Slides] Lecture 10 video lecture107 min [LLaVA], [Moshi], [OpenVLA], [NExT-GPT] Assignment 3 due (11:59pm)
Dec. 4th Lecture 11: Efficiency in LLMs FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness
Efficient Memory Management for LLM Serving with PagedAttention (vLLM)
Fast Inference from Transformers via Speculative Decoding
Switch Transformers: Scaling to Trillion Parameter Models
LoRA: Low-Rank Adaptation of Large Language Models
QLoRA: Efficient Finetuning of Quantized LLMs
ZeRO: Memory Optimizations Toward Training Trillion Parameter Models
[Slides] Lecture 11 video lecture147 min [vLLM], [FlashAttention], [PEFT], [DeepSpeed]
Dec. 11th Lecture 12: Guest Lecture / LLM Applications and the Future Survey of Hallucination in Natural Language Generation
Introducing Superalignment
GPTs are GPTs: Labor Market Impact Potential of LLMs
Dec. 18th Final Project Presentation Final Project Poster due (17:00pm Dec. 17th)
Dec. 25th Final Report Submission Final Report Submission due (11:59 pm Dec. 25th)

Freedom Intelligence: Research, Innovation & Startup Ecosystem


This course is connected to Freedom Intelligence, an AI research and innovation team based at CUHK-Shenzhen, SRIBD and Shenzhen Loop Area Institute. The team works across large language models, medical AI, multimodal intelligence, agents, evaluation and AI for society, translating frontier research into open-source systems and real-world applications.

Strong innovation and startup pathway: students learn in an ecosystem with strong GPU and engineering resources, close collaboration with hospitals and industry, and experience spanning research prototypes, open-source adoption, product validation and deployment. Freedom Intelligence reports 10K+ GitHub stars, 1M+ Hugging Face downloads, 300K+ HuatuoGPT service visits and 520K+ clinical AI interactions. Its alumni network also includes the co-founder of Cognita, an AI startup acquired by Mosaic Clinical Technologies.

Visit Freedom Intelligence Meet the Team GitHub Hugging Face

Course Information


This comprehensive course on Natural Language Processing (NLP) offers a deep dive into the field, providing students with the knowledge and skills to understand, design, and implement NLP systems. Starting with an overview of NLP and foundational linguistic concepts, the course moves on to word representation and language modeling, essential for understanding text data. It explores how deep learning, from basic neural networks to advanced transformer models, has revolutionized NLP and its diverse applications, such as text mining, information extraction, and machine translation. The course emphasizes large language models (LLMs), their scaling laws, emergent abilities, training strategies, and associated knowledge representation and reasoning. Students will apply their learning in final projects, for example, exploring NLP beyond text with multi-modal LLMs, AI for Science, vertical applications and agents. There are guest lectures and in-class paper discussions that could learn the cut-edge research. The course also concludes with an examination of NLP's limitations and ethical considerations. In particular, the topics include:

Prerequisites

Learning Outcomes

Grading Policy

Assignments (40%)

Final project (55%)

The project could be done by a group but each indivisual is separately evaluated. You need to write a project report (max 6 pages) for the final project. Here is the report template. The Final Project Guide & Challenge Tracks page (English and Chinese) lists eight project tracks, 77 featured briefs, competitions and resources; the writing tips page collects the Lecture 8 writing advice with poster and paper samples. You are also expected to make a project poster presentation. After the final project deadline, feel free to make your project open source; we appreciate if you acknowledge this course

Optional: You may submit a project proposal when you build up your team for feedback before the final deadline.

Participation (5%)

Here are some ways to earn the participation credit, which is capped at 5%.

Late Policy

The penalty is 20% off the assignment grade for each late day.


Coursework at a glance

Everything you hand in this term, with the pages that explain each piece. Assignments 40% · final project 55% · participation 5%.

55%

🎯 Final project

A four-week research study in a team of 2–3: one question, one baseline, one targeted change. Eight tracks and 79 briefs to choose from, in English and Chinese.

hands-on

💻 Tutorials

Friday sessions that turn the lecture into code: notebooks you can open in Colab, with slides and a self-check quiz on each tutorial page.

40%

📝 Assignments

Three individual assignments, each with a Colab notebook or a short report. Submit on Blackboard by 11:59 pm (Shenzhen time) on the due date.

Oct. 23 Assignment 1 due Nov. 6 Assignment 2 due Nov. 13 project teams & optional proposal Dec. 4 Assignment 3 due Dec. 17 poster due 17:00 Dec. 18 poster session Dec. 25 final report due

Past Offerings

Spring 2026  |  Fall 2025  |  Spring 2025  |  2024  |  2023


Acknowledgement

We borrowed some concepts and the website template from [CSC3160/MDS6002] where Prof. Zhizheng Wu is the instructor.

Website github repo is [here] .