[2025 Fall] Review

It was challenging semester for me. I’m happy that I was able to finish everything on time, I’m not sure I did everything very well though.

cmpt 464[dropped] - Richard Zhang

This course is not a classic geometry processing course. It’s more about 3d vision or 3d ML. Dr.Zhang gives a good insight on the topic or even outside of topics, graphics in general. I like the course content, but I had to drop it because of the workload of capstone and graduate course. I wish I could have another chance on this.

cmpt 307

This was a mandatory algorithm analysis course. The design was somewhat challenging, with four quizzes and a final exam with no graded assignments, but it helped me focus on one topic at a time. I wouldn’t say I’m deeply interested in analyzing algorithms for efficiency and correctness, but learning something new always offers fresh perspectives. It helped me think about optimization and its connection to machine learning “algorithms.” There’s always a gap between fundamentals and cutting-edge implementations, which is both the strength and limitation of studying foundational topics.

cmpt 495 - Ke Li

The capstone is done. This semester we worked in a team of three with separate tasks. Mine was porting the backend inference code to run in the browser. I used WebGPU-Torch, a WebGPU implementation of PyTorch, but it stopped being maintained about two years ago and lacks some functions. That made things challenging, but we managed well and produced solid insights and results. I mostly focused on getting the PAPR model running in the browser, implementing custom kernels missing from WebGPU-Torch, and organizing the project repo. Looking back, I spent about six months on this capstone and learned a lot. I’m planning to continue in this direction, so there’s more to come.

cmpt 983 - Ke Li

This is a graduate-level course on generative modeling, covering everything from foundational algorithms like EM to recent models like diffusion. I found it challenging, but Dr. Ke Li’s courses are always satisfying because of the well-structured content and how organically the materials connect. The final project was also a good introduction to how graduate courses work. I’d like to revisit and deepen my understanding when I have time.