🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

We need more of *Science of Deep Learning* in the major ML conferences. This year’s NeurIPS Conference workshop Scientific Methods for Understanding Deep Learning on this topic is just starting, and I hope it is NOT the last edition!!!

Now onto Surya Ganguli talk, investigating the creativity in diffusion models, i.e. why do diffusion models fail to learn the score function and why it's a good thing.

Excited for the NeurIPS workshops today! I'll be speaking at: (1) Science for DL (panel, 3:10-4:10, scienceofdlworkshop.github.io/schedule/) (2) "Time Series in the Age of Large Models" (talk, 4:39-5:14, neurips-time-series-workshop.github.io).

Can LLMs plan? The answer might be in Hanie Sedghi's talk. Only one way to know, join us in West meeting rooms 205-207

Absolutely!

We are back with a contributed talk from Christos Perivolaropoulos about some intriguing properties of sotfmax in Transformer architectures!

Moving onto David Krueger's talk about Input Space Mode Connectivity in Deep Neural Networks! Come learn about how you can interpolate between loss minimizers.

Poster session still going on in West meeting rooms 205-207! Come say hi and check your favorite posters (also there is decaf)

Hopfield models are hot again to initialize memorization and generalization in diffusion models in Bao Pham contributed talk! 🧐

Last talk of the day (before our panel session), Mikhail Belkin shares his personal experience and his learnings from experiments (in an alchemist way? 😎)

It's time for our panel session! Surya Ganguli Eero Simoncelli Andrew Gordon Wilson Yasaman Bahri and Mikhail Belkin discuss science, understanding and deep learning. 👀 Hosted by our amazing organizers Zahra Kadkhodaie and Florentin Guth.

What a great panel Scientific Methods for Understanding Deep Learning moderated by Zahra Kadkhodaie and Florentin Guth ! #NeurIPS2024

Congratulation to the winners of our debunking challenge François Charton and Julia Kempe for "Emergent properties with repeated examples" 🥳🎉

One epoch is not all you need! Our paper, Emergent properties with repeated examples, with Julia Kempe, won the NeurIPS24 Debunking Challenge, organized by the Science for Deep Learning workshop, Scientific Methods for Understanding Deep Learning arxiv.org/abs/2410.07041