Tokyo Tech Talks #8: Journey to the Centre of the LLM
Larry Richards will guide us inside a large language model to explore what happens beneath the familiar surface of prompts and responses.
We’re excited to announce Tokyo Tech Talks #8: Journey to the Centre of the LLM.
We usually encounter a large language model from the outside. We provide words, the model returns words, and its internal activity remains largely hidden from view. What might we learn if we could venture inside?
The Speaker
Larry Richards
Larry is an independent researcher studying mechanistic interpretability, sparse representations, activation interventions, and the geometry of language-model internals.
He is also the creator and maintainer of vollib, an open-source library with Python and TypeScript implementations that provide the same functionality for option pricing, implied volatility, and Greeks. Larry publishes his research and experimental notes at Larry Richards - Mechanistic Interpretability Research.
The Talk: Journey to the Centre of the LLM
Larry will guide us on a journey into the internal behavior of a large language model, exploring what happens beneath the familiar surface of prompts and responses.
Who Should Attend
This talk is for people who already have a basic understanding of Transformer models, including layers, activations, and next-token prediction. It will explore model internals experimentally rather than introduce Transformer architecture from the beginning.
For anyone who would like to prepare, 3Blue1Brown’s But what is a GPT? Visual intro to transformers offers an excellent foundation.
Event Details
- Date: Tuesday, September 29, 2026
- Time: 7:00 PM - 9:00 PM JST
- Location: Build+, 6F, MARIX Ebisu Bldg., 4-4-6 Ebisu, Shibuya, Tokyo
- View on Google Maps
Schedule
- 7:00 PM - 8:00 PM: Talks
- 8:00 PM - 9:00 PM: Drinks & Networking
Join Us
Follow the Tokyo Tech Talks Luma calendar for registration updates.
Come venture inside the LLM with us, then stay for drinks and conversation.