Henok | Neural Nets
1.61K subscribers
233 photos
20 videos
13 files
157 links
Download Telegram
Attending Neurips has always been in my list and will continue to be. If you like ML research this is a great one. I also presented my paper there, attended hundreds of posters. And so many people like Yoshua Bengio, Fei-Fei Li, François Chollet, Ilya Sutskever, Yann LaCun, Noam Brown were here and you can just walk by one of them.

Next year it will be in the 🌞 sunny San Diego, California. I'm hoping to be there and probably I'll see one of you :).

And finally Vancouver is the best. Discovered cool places with @pi_314159265 and he's simply an amazing guy, learned lots of things on the way too.

And some random pictures
🔥218
So last week I met someone in a mixer and we were chatting about some things and guess what, today I found out he is D Sculley, CEO of Kaggle, damn.
👏17🔥144😱1
Media is too big
VIEW IN TELEGRAM
🤯🤯🤯 what

Genesis, a generative physics engine which can generate 4D dynamical worlds.

https://x.com/zhou_xian_/status/1869511650782658846
🔥10🤯6
So they built an agent and then set it to work on its own repo. Now, it's the most active contributor on their team.

I can't believe that less than 8 months ago OpenHands
(then, OpenDevin) sent its first pull request, and now it's the most active contributor to its own codebase: https://github.com/All-Hands-AI/

We're living in amazing times.



Source
6😁3🌚3
One of the things I did in Vancouver was hangout with Joseph and Israel.

Joseph in the 90s has listed his company on Nasdaq, probably the first and only Ethiopian and then later sold the company for more than $100m.

You guys know Aman, a great researcher and Chapa CTO, Israel.
🔥1643👍1
Is AGI possible in the next 200 years?
Media is too big
VIEW IN TELEGRAM
This is a poster session and this is almost 1/10 of the posters there. Just to give you a glimpse.


Almost 15k people attended Neurips and everyone is working on different areas of ML, what's the ultimate goal of their work and in 25 or 50 years most of the tasks they are trying to do will be solved, they will solve it, i think.

Now all those little things combined, pose estimation, inference time, better architectures, recognition, classification etc are just small pieces to really make AI!?

Oh and academia isn't dead but not all the papers here are relevant 😂
👍4
Apply to learn Computational Neuroscience and Machine Learning in Africa.

7-23th July 2025,
University of Zambia, Lusaka, Zambia

An intensive entry level course to teach young African students and researchers the basics of computational neuroscience and machine learning


https://trendinafrica.org/trend-camina
👍54
"For those unfamiliar, an LSTM is the thing that poor deep learning researchers did before transformers, and it's basically a ResNet, but rotated 90 degrees"

Ilya



I need this on a shirt, simply the most artistic way of talking about DL
5👍1