Muawiz Chaudhary

🌈 I'll be presenting our JMLR paper "A rainbow in deep network black boxes" today at 3pm at @iclr-conf.bsky.social!
Come to poster #334 if you're interested, I'll be happy to chat
More details in the threads on the other website: x.com/FlorentinGut...

Having a lot of fun presenting our work with Hadi on RL for microgrid optimization with industrial constraints at Climate Change AI workshop at NeurIPS!

If you are at NeurIPS, come and visit us during the poster session :)

@hnekoei.bsky.social
@climatechangeai.bsky.social

Discover the Evaluating Evaluations: Examining Best Practices for Measuring Broader Impacts of Generative AI workshop, co-organized by Mila student @mchll-ln.bsky.social at @neuripsconf.bsky.social, East Meeting Room 16.

📢 New paper “Language model scaling laws and zero-sum learning” at Sci4DL #neurips2024.

ℹ️ openreview.net/forum?id=yBq2g832Go TL;DR: scaling improves LMs by mitigating zero-sum learning, a mechanism that could be targeted directly and independent of scale.

West 205-207 4:30-5:30 PM

🧵 (1/12)

this talk by @suryaganguli.bsky.social at #Sci4DL is makin me think I should have applied for Applied Physics PhDs instead of Comp Sci PhDs.

@mila-quebec.bsky.social

My collaborators (Vivian White, @kamdh.bsky.social) will be presenting our work at the #Sci4DL workshop at #NeurIPS2024 today.

Location: West Meeting Room 205-207
Time: 4:30-5:30 PM

We present a principled probability distribution model of pre-trained deep neural networks. Check it out!

Learning Stochastic Rainbow Networks Random feature models are a popular approach for studying network learning that can capture important behaviors while remaining simpler than traditional training. Guth et al. [2024] introduced...

Check out also our work which investigates holding the initialization of a convolution fixed while learning a structured covariance part.

Our 2nd paper uses this to sample trained DNNs and investigate eigenvalue distributions.

openreview.net/forum?id=bvJ...
openreview.net/forum?id=vWh...

A similar result is found in "A Rainbow in Deep Network Black Boxes". The initialization of networks stays the same over training and whats learned is the covariance. See figure 10 and pages 33-36. arxiv.org/pdf/2305.18512 @florentinguth.bsky.social

LeMaterial: an open source initiative to accelerate materials discovery and research We’re on a journey to advance and democratize artificial intelligence through open source and open science.

⚛️🤗 Announcing LeMaterial ⚛️🤗

@hf.co & @entalpic.bsky.social are teaming up to release LeMaterial -- an open source initiative aiming to facilitate (AI for) materials discovery !

Datasets, hash function, tools to explore the chemical space & more !

huggingface.co/blog/lemater...

I want my grad school apps to be done with and go back to focusing on research.

As AI researchers, we shouldn’t demonize people outside our space who have a passionate distaste for AI. You have to understand that most of the pro-AI sentiment people see online comes from absolutely vile “AI-bros”, especially on twitter. We just need to distinguish ourselves as academics.

I'd like to be added :)

always happy to chat about this work

If your interested in what I've been working on the past year, heres a talk by @kamdh.bsky.social my senior collaborator discussing our progress in the rainbow hypothesis line of work :)