Stanford CS336: Language Modeling from Scratch (on-going)

www.youtube.com/playlist?lis...

Generative modelling in latent space Latent representations for generative models.

New blog post: let's talk about latents!
sander.ai/2025/04/15/l...

We often use discretization to approximate continuous laws of physics, but it also goes the other way:

You can use continuous equations to approximate the behavior of discrete systems!

Here we'll see how electrical circuits can be modeled using the Laplace equation Δφ=0. [1/n]

Foundations of Computer Vision

Just found out that Foundations of Computer Vision is now available online! 📚Definitely worth checking out—I really enjoyed it.

📢PSA: #NeurIPS2024 recordings are now publicly available!

The workshops always have tons of interesting things on at once, so the FOMO is real😵‍💫 Luckily it's all recorded, so I've been catching up on what I missed.

Thread below with some personal highlights🧵

Great way to expand one's graphics knowledge: Every day grab and read a random graphics paper from realtimerendering.com/random.html. Even if you struggle to get the specifics, read the abstract or skim over it to learn about its existence and get the gist of a graphics technique.

Casual Viewing | Will Tavlin A decade before Airbnb persuaded homeowners to transform their homes into hotels, Netflix convinced its users to turn theirs into mini Netflix warehouses. Customers who held onto their DVDs for longer...

This deep dive on Netflix purposely turning movies into barely distinguishable forgettable slop is so insightful, researched & brutally hilarious. A must read if you want to really know what’s happening to the business AND culture of movies.

> No one says "why should only a few people get to run Marathons." If you want more people to run marathons, you don't distribute mopeds!

Summary of Ted Chiang’s NeurIPS talk about “GenAI and Art.” x.com/rao2z/status...

Entropy is one of those formulas that many of us learn, swallow whole, and even use regularly without really understanding.

(E.g., where does that “log” come from? Are there other possible formulas?)

Yet there's an intuitive & almost inevitable way to arrive at this expression.

A common question nowadays: Which is better, diffusion or flow matching? 🤔

Our answer: They’re two sides of the same coin. We wrote a blog post to show how diffusion models and Gaussian flow matching are equivalent. That’s great: It means you can use them interchangeably.

Check out CAT4D: our new paper that turns (text, sparse images, videos) => (dynamic 3D scenes)!

I can't get over how cool the interactive demo is.

Try it out for yourself on the project page: cat-4d.github.io

Anne Gagneux, Ségolène Martin, @quentinbertrand.bsky.social Remi Emonet and I wrote a tutorial blog post on flow matching: dl.heeere.com/conditional-... with lots of illustrations and intuition!

We got this idea after their cool work on improving Plug and Play with FM: arxiv.org/abs/2410.02423

A great way to convert your X network on Bluesky. It worked very well for me.