Cover Image for The Toronto LLM Meetup Group - Model Merging
Cover Image for The Toronto LLM Meetup Group - Model Merging
79 Went

The Toronto LLM Meetup Group - Model Merging

Hosted by The Toronto LLM Meetup Group, Suhas Pai & Lily (Xiaojun) Ren
Registration
Past Event
Please click on the button below to join the waitlist. You will be notified if additional spots become available.
About Event

Welcome to the 6th session of the 4th edition of the Toronto LLM Meetup Group. We really appreciate the growing interest in our events and will continue to ensure equitable access for all participants. This time, we’ll be exploring Model Merging.

2023-2024 were the heydays of model merging, when we were witness to thousands of artisanal merges appearing on Hugging Face, some of them reaching dizzying heights up the eval leaderboards. The year of agents and Clawdbots have somewhat dampened enthusiasm for this esoteric field, but we think this area is ripe for a renaissance. 

In this session, we will provide a highly technical, in-depth explanation of why model merging works, the various ways merging can be done, and some new adjacent research directions. We will also showcase cool demos that hint at the limits of what is possible.

​​Special thanks to Ada for providing the event space and Hudson Labs for their support and resources.

🧧🌙✨ Note that we will be serving halal snacks and beverages to celebrate CNY & the possible first day of Ramadan as many of you might be breaking fast during the event


​📖 Papers & Resources

​​It is not necessary to read them all before the session, but browsing through them might help you more actively participate in the discussion session.

https://arxiv.org/pdf/2408.07057 - A survey on model merging

https://arxiv.org/pdf/2510.09913 - Don't throw away your pretrained model

https://arxiv.org/pdf/2510.04686 - How does the optimizer implicitly bias the model merging landscape?


​​⏱️ Timetable

​​17:30 - 18:00 Icebreaker & Networking

​​18:00 - 18:30 Speaker Session #1

​​18:30 - 18:50 Discussion

​​18:50 - 19:20 Speaker Session #2

​​19:20 - 19:40 Discussion

​​19:40 - 20:00 Networking


​​📣 Speaker Bios

Manav Shah is a Founding Engineer and Head of Engineering at Impel Intellingence — makers of Dippy AI. He is experienced in building and scaling consumer products, engineering systems for rapid iteration and designing RLHF pipelines. He previously worked on RL and Recommender Systems at RBC Capital Markets.

​​More to come...


​​👋 About Us

​​The Toronto LLM Meetup group is founded by Suhas Pai, CTO and Co-founder of Hudson Labs, and Lily Ren, Senior PM, B2B AI at Telus. The aim of this group is to facilitate deep dives into LLM research topics, harnessing the skills of Toronto's vibrant AI ecosystem and promoting intellectual rigor. To this end, for each session we assemble a team of top Toronto AI scientists and practitioners to conduct an in-depth study of an emerging topic that is expected to go mainstream 6 months from now. The team presents their findings along with jumping-off points for future research at a meetup session, held at Toronto's top AI companies. Eventually we aim to spawn off community open-source research projects this way.


​​📧​ Mailing List

​​Thank you so much for your interest. We’re thrilled to see so many of you sign up for this event! We’ll be collecting emails to invite you to future editions of the meetup, but if you’d prefer not to be on the mailing list, just let us know.

Location
46 Spadina Ave. 5th floor
Toronto, ON M5V 2H8, Canada
79 Went