​For this guest speaker session, we'll have Sarath Chandar (Canada CIFAR AI Chair) visiting Tokyo and joining us for a talk on LLMs. In this talk, Sarath will give an overview of several recent projects from his lab on developing efficient and safe large language models. First, he will introduce NeoBERT, a new state-of-the-art encoder model for text. Then he will talk about the benefits of encoder models over decoder models for specific tasks. Then, Sarath will show how we can improve the reasoning capabilities of LLMs by better packing during training. For the rest of the talk, we will discuss how to make LLMs safer through various techniques, including model merging and steering.

​Agenda

​18:30 Doors open

​19:00 - 20:00 Developing Efficient Large Language Models (Sarath Chandar)

​20:00 - 20:45 Networking

​21:00 Doors close

​Bio

​Sarath Chandar (Associate Professor at Polytechnique Montreal and Mila)

​Sarath Chandar is an Associate Professor at Polytechnique Montreal, where he leads the Chandar Research Lab. He is also a core faculty member at Mila, the Quebec AI Institute. Sarath holds a Canada CIFAR AI Chair and the Canada Research Chair in Lifelong Machine Learning. His research interests include lifelong learning, deep learning, optimization, reinforcement learning, natural language processing, and AI for science. To promote research in lifelong learning, Sarath created the Conference on Lifelong Learning Agents (CoLLAs) in 2022 and served as a program chair for 2022 and 2023.

​​​​​​​Tokyo AI (TAI) information

​​​​TAI is the biggest AI community in Japan, with 2,400+ members mainly based in Tokyo (engineers, researchers, investors, product managers, and corporate innovation managers).

​Web: https://www.tokyoai.jp/