The Toronto ML Systems Group: Diffusion Models
Come hang out with The Toronto ML Systems Group. The theme for our fourth event is diffusion models.
We’re bringing together people who care about the full stack behind modern AI, from chips and kernels to ML compilers, model architectures, attention variants, distributed training, inference optimization, and the hardware-software tradeoffs that shape what’s possible.
This meetup is meant to be educational, recurring, and a little more conversational than formal, with enough structure to help you follow the whole stack without getting lost in the weeds. Whether you’re already working in ML systems or just want a clearer path into the space, expect good conversation, practical ideas, and plenty of room to dig into how these systems are built.
Tentative schedule
6:00 – 6:30 PM Check-in, food, and networking
6:30 – 6:35 PM Welcome, intros, and agenda
6:35 – 6:50 PM Ideogram: inside the Ideogram 4 architecture
6:50 – 7:00 PM Discussion and Q&A
7:00 – 7:20 PM Transformer Lab : fused kernels for Ideogram 4 and quantizing it to run on a single 4090
7:20 – 7:30 PM Discussion and Q&A
7:30 – 8:00 PM Open floor: attendees share what they’re working on related to the theme of diffusion models.