0%
COMP5046 · NLP · S1 2026
Natural Language Processing
Interactive study hub aligned with lecture handouts: guides, foundations, mind maps, quizzes, and flashcards — with bilingual UI.
Coverage
11 structured lectures
Toolkit
Study guides, maths, quizzes & flashcards
Format
Handout + workshop notebook synthesis
Mission control
Turn your study hub into a mastery loop: keep momentum, see progress, and jump straight to the right lecture.
Awaiting first move
0
No learning streak yet. Open a lecture and complete a check to start one.
0 / 6
Clear a lecture by building strong quiz, quick-check, and study coverage.
Start Lecture 1
Use the study guide, then jump into foundations and quizzes once the basics are set.
Launch mission →Lecture atlas
Choose a lecture, then move from core ideas to foundations, quizzes, and revision tools.
Lecture 1
Introduction & Representing Text
Course goals, WordNet, one-hot & bag-of-words, TF–IDF & BM25, distributional similarity, CBOW/word2vec preview, and embedding evaluation.
Ready
EN / 中文
Lecture 2
Foundations of NLP Systems
Five components, sentiment baseline, BoW features, linear models, embeddings, train/dev/test, precision/recall/F1, ROC.
Ready
EN / 中文
Lecture 3
Non-linear Models
MLP activations, backpropagation, regularisation, RNNs, transducer vs encoder, gradient vanishing/exploding.
Ready
EN / 中文
Lecture 4
Inference: Greedy & Search
Exhaustive vs greedy, beam search, sampling variants, graph search, sequence tagging & Viterbi.
Ready
EN / 中文
Lecture 5
Encoder–Decoder Models
Static & contextual embeddings, seq2seq, MT eval (chrF, BLEU), BPE, bottleneck to attention.
Ready
EN / 中文
Lecture 6
Transformer Models
Attention variants, self-attention Q/K/V, positional encodings, masking, multi-head, encoder–decoder.
Ready
EN / 中文
Lecture 7
Language Models
N-gram models, Markov assumption, smoothing, perplexity, and neural language models.
Ready
EN / 中文
Lecture 8
Data & Annotation
Data sources (Common Crawl, Wikipedia), annotation, crowdsourcing, inter-annotator agreement, risks.
Ready
EN / 中文
Lecture 9
Training: PEFT & LoRA
Pre-training vs fine-tuning, parameter-efficient fine-tuning, LoRA / QLoRA, adapters.
Ready
EN / 中文
Lecture 10
Training: RLHF
Preference data, reward models, PPO with KL penalty, DPO alternative, alignment risks.
Ready
EN / 中文
Lecture 12
LLM Agents
Chain-of-Thought reasoning, ReAct, retrieval-augmented generation, tool use, and risks.
Ready
EN / 中文