Suchir Teaching
Teaching & Supervision
Guest lecturer, teaching assistant, and supervisor across the Cambridge Computer Science and Linguistics Tripos.
Guest Lecturing
Cambridge Lecture Notes
I am a guest lecturer for several courses in the Linguistics and Computer Science Triposes. Here you can find my lecture materials, handouts and associated repositories.
I delivered my first guest lecture in November 2024 for an MPhil course at the University of Cambridge, with Prof Paula Buttery and Dr Fermín Moscoso del Prado Martín, on Language Model Evaluation. Since then I have contributed lectures and handouts on language modelling, parsing, and evaluation across the graduate and undergraduate curriculum.
L95 — Introduction to Natural Language Syntax & Parsing
- Part III / MPhil Advanced Computer Science
- Guest lecturer and teaching assistant. My handout covers Language Model Evaluation (BLiMP and minimal-pair benchmarks), delivered in Michaelmas 2024.
- Course topics include parsing algorithms (CKY, shift-reduce), categorial grammars, span-based constituency parsing, evaluation, lexicalised PCFGs and unification grammars, and recent topics such as discontinuous parsing.
Li18 — Computational Linguistics
- Linguistics Tripos
- 2025–26 lecture notes on Computational Linguistics — Language Models and Parsing, and Deep Neural Networks.
- Introduces students to statistical and neural approaches to language, from n-gram and PCFG parsing through to modern transformer language models.
Supervisions
Cambridge Supervision Resources
I supervise several courses for Cambridge Computer Science undergraduates, and am collating lecture notes, handouts and supervision materials for courses I supervise or have previously taken.
CST IA — First Year
- Introduction to Probability
- Machine Learning & Real World Data
- Role: Supervisor & Teaching Assistant / Lead Demonstrator
CST IB — Second Year
- Artificial Intelligence
- Formal Models of Language
- Role: Supervisor
CST II — Third Year
- Machine Learning & Bayesian Inference (MLBI)
- Role: Supervisor
- Supervision resources cover: MAP Learning (regression & classification); optimum weight vector for Ridge Regression and Iterated Reweighted Least Squares (IRLS); SVMs and KKT conditions; and the Bayesian interpretation of neural networks (Matrix Inversion Lemma).
UROP Supervision
Small Language Models UROPs 2025
An eight-week Undergraduate Research Opportunity Programme built on Pico, our small language model learning-dynamics framework, co-supervised with Prof Paula Buttery.
In 2025 I co-supervised two Small Language Model UROP students, running a structured practical course alongside their independent research projects. The programme pairs a weekly reading list with hands-on pretraining and interpretability work using Pico. I am also a Google DeepMind Research Ready mentor together with Richard Diehl Martinez and Prof Paula Buttery, supported by Google DeepMind, the Hg Foundation and the Royal Academy of Engineering.
Eight-Week Syllabus
Week 1 (14 Jul) — BabyLMs and Multilingual Evaluation
- Required: Findings of the 1st BabyLM Challenge; Findings of the 2nd BabyLM Challenge; Handout on LM Evaluation.
- Recommended: CLIMB; Less is More (cross-lingual curriculum).
Week 2 (21 Jul) — Tokenisation & Interpretability / Bilingual BabyLM
- Linear Representation Hypothesis; Arthur Conmy (DeepMind) slides; Universal Dependencies.
- Shared Grammatical Representations in Bilingual LMs; MultiBLiMP 1.0; TransformerLens.
Week 3 (28 Jul) — Pretraining LMs: The Pico Framework
Week 4 (04 Aug) — BabyLM Architectures & Feedback (ALTA, CST)
Week 5 (11 Aug) — Mechanistic and Developmental Interpretability
- Interim project presentations.
Week 6 (18 Aug) — Train Your Own BabyLM From Scratch
- Interaction Track / Multimodal Track.
Week 7 (25 Aug) — Small Language Models: Frontier Problems
- Linear Representation Hypothesis; Conmy slides.
Week 8 (01 Sep) — Small Language Models: Frontier Problems (Architectures)
- Final project presentations. UROP reports due Friday 5 September 2025, 5pm.
Language Model Primers
- CS336 — Language Models from Scratch; CS224U — Contextual Word Representations; CS224U — In-Context Learning.
Small Language Models & Further Reading
Project Supervision
MPhil Projects & Supervised Research
Project proposals for prospective MPhil students, together with materials from projects and research reports I have supervised.
MPhil Project Proposals 2025–26
- Proposed MPhil (Advanced Computer Science / MLMI) project topics in small language models, multilinguality and interpretability.
Supervised Projects & Reports
- Selected reports and write-ups from MPhil theses, UROP projects and research projects I have supervised or co-authored.
Tripos Advice
- A short set of notes for Computer Science and Linguistics Tripos students on approaching Natural Language Processing coursework, projects and supervisions.