NLP & LLMs (Spring 2026)
Fudan University · CS40008.01 · Location: HGX104 · Thursdays 1:30–4:10pm, 5 March – 18 June 2026
Full course site: baojian.github.io/llm-26 — slides, notes, and assignments live there.
Syllabus
| Week | Topic |
|---|---|
| 1 | Introduction to LLMs — tokenization (BPE/WordPiece), vocabulary design |
| 2 | N-gram language models — MLE, smoothing, perplexity |
| 3 | Word embeddings — the distributional hypothesis, Word2Vec, text classification |
| 4 | Neural language models — sequence learning, LSTM, encoder–decoder |
| 5–6 | Attention and the Transformer |
| 7 | LLM pretraining (GPT) |
| 8 | Evaluation and benchmarks |
| 9 | BERT and post-training — bidirectional encoders, masked language modelling |
| 10 | Post-training — SFT, reward modelling, PPO/RLHF |
| 11 | Information retrieval and retrieval-augmented generation |
| 12 | Course project presentations |
| 13 | Diffusion language models |
| 14 | Alignment and safety |
| 15 | Efficiency and systems |
| 16 | Agents and frontiers |