Me: Hey, this research idea seems simple enough and pretty interesting. I'll allocate a week and try it real quick. 3 days later Me: Huh, so that's how kubernetes work...

Anyone knows where are the demos? @cvprconference.bsky.social

They say the second is the first loser... Happy to announce we got second best paper in the Event Vision workshop @cvprconference.bsky.social for the paper Probabilistic Online Event Downsampling! www.arxiv.org/pdf/2506.02547

At #CVPR2025! So far: + Breakfast was super good! - While tutorials and workshops said "from 8" in their sites, it seems that the official time has been delayed to "from 9" without any update or contact from the organizers. So yep, here we are, the ones who woke up early and ate fast, waiting.

Deadline mode

CVPR is getting savage, all my reviewer batch rejected :0

Reading some reviews and the corresponding rebuttals is like looking at a conversation between two chat gpts...

Great talk by @abursuc.bsky.social on foundational models, fine tunnings and the lots of difficulties on autonomous driving #MMM2025

One of the best parts of any conference #MMM2025

First keynote by Nancy Chen begins! #MMM2025

Ready to go! #MMM2025

Getting ready! #MMM2025

Omw to finish the preparations for #MMM2025 in Nara! My first experience in an international conference organizing committee, it will be a blast!

I'm always conflicted with the science (?) of AI research. Sometimes I read "we discovered that doing A is useful to solving this task", but in my mind is "we made A, and forced it to solve this task by tweaking this and that". Am I missing something?

4 papers assigned for #CVPR25 and all super well aligned with my research. Good job ACs! @cvprconference.bsky.social

I made a meme

When working with docker + live cameras is latency super noticeable?

Registration for #DLBCN 2024 is now open for the general public. Get your ticket today ! sites.google.com/view/dlbcn20...

Incredible work being done by @drfeifei.bsky.social et al. on 3D world generation. Also super cool demos! www.worldlabs.ai/blog

Looking at ICLR submissions with the lowest score - What a work of art! 🧵

My deep learning course at the University of Geneva is available on-line. 1000+ slides, ~20h of screen-casts. Full of examples in PyTorch. fleuret.org/dlc/ And my "Little Book of Deep Learning" is available as a phone-formatted pdf (nearing 700k downloads!) fleuret.org/lbdl/

Piggybacking on this thread, I have the feeling that authors tend to maximize not only on the quality of the research, but also on how to improve the chances of being accepted. This usually means not showing super interesting insights that would "weaken" their method, and I'm kind of sad for it.

Transformer by hand ✍️ in Excel by Tom Yeh He has released his first-ever "Full-Stack" implementation of the Transformer model. Repo: github.com/ImagineAILab...

What do people use these days to track/log experiments? Asking for a friend.

I love the starter packs people are making. Somebody, please, make the "Academia memelord" starter pack!

Well, let's try again.