Scientific Methods for Understanding Deep Learning
@scifordl
Followers
349
Following
21
Media
18
Statuses
39
Workshop @ NeurIPS 2024. Using controlled experiments to test hypotheses about the inner workings of deep networks.
Joined July 2024
RT @f_charton: One epoch is not all you need! Our paper, Emergent properties with repeated examples, with.@KempeLab, won the NeurIPS24 Debu….
0
17
0
Congratulation to the winners of our debunking challenge @f_charton and @KempeLab for "Emergent properties with repeated examples" 🥳🎉
0
4
33
RT @SharvVadgama: What a great panel @scifordl moderated by @ZKadkhodaie and @FlorentinGuth !. #NeurIPS2024
0
2
0
It's time for our panel session! @SuryaGanguli @EeroSimoncelli @andrewgwils @yasamanbb and Mikhail Belkin discuss science, understanding and deep learning. 👀 Hosted by our amazing organizers @ZKadkhodaie and @FlorentinGuth.
1
5
30
Last talk of the day (before our panel session), Mikhail Belkin shares his personal experience and his learnings from experiments (in an alchemist way? 😎)
0
2
14
"Are we really doing science?", "What are we really doing as scientists?", @zicokolter with a 🔥 lukewarm 🔥 defense of alchemy in Deep Learning (including some hot (?) takes).
1
1
10
Hopfield models are hot again to initialize memorization and generalization in diffusion models in @baophamhq contributed talk! 🧐
0
1
8
After a well-deserved (?) lunch break, we are back at it with the contributed talk of Antonio Sclocchi unraveling the hierarchical structure of data using diffusion models!
1
1
12
Poster session still going on in West meeting rooms 205-207! Come say hi and check your favorite posters (also there is decaf)
0
1
2
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.
0
0
9
We are back with a contributed talk from @ccperivol about some intriguing properties of sotfmax in Transformer architectures!
0
5
28
Can LLMs plan? The answer might be in @HanieSedghi's talk. Only one way to know, join us in West meeting rooms 205-207
0
3
8
RT @andrewgwils: Excited for the NeurIPS workshops today! I'll be speaking at:.(1) Science for DL (panel, 3:10-4:10, .
0
6
0
RT @zdeborova: We need more of *Science of Deep Learning* in the major ML conferences. This year’s @NeurIPSConf workshop @scifordl on this….
0
16
0
Now onto @SuryaGanguli 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.
1
3
11
@tomgoldsteincs Discussing weak-to-strong generalization: "I want a net that solves harder problems than it was trained on"
1
1
3