sparsenn

Sparsity in Neural Networks Workshop

May 5th 2023

[sparsenn @sparsenn]

📢The call for papers of the “Sparsity in Neural Networks: On practical limitations and tradeoffs between sustainability and efficiency” Workshop is online at https://t.co/lYkj8mkRYY.
🗓️Deadline: February 3, 2023, AoE.
#ICLR2023

[sparsenn @sparsenn]

🙏 for joining us at 7am PT!

[Jeff Dean @JeffDean]

@stanfordnlp @Google @DeepMind Our presence at last week's ICLR seems to be pretty strong evidence against this, given 100+ papers with Google co-authors we published there, and numerous conference organizing and workshop roles/talks (I gave one myself in the @sparsenn workshop).

https://t.co/5QYkHA9Lic

[Gintare Karolina Dziugaite @gkdziugaite]

Check out a new pruning library, JaxPruner! Excited to see how it will impact those already working in network pruning and quantization, and attract new people interested in trying out / applying these methods in new domains 🚀

[Kyunghyun Cho @kchonyc]

before and after #iclr2023

[Vithu Thangarasa @vithursant19]

@white_martha Brought up a great point during the panel discussion at the @sparsenn workshop, “rather than taking existing architectures, which we’ve been working with for many years are kind of designed for dense architectures, and then saying...that’s our gold standard...(1/2)

[Thiago Serra (@thserra.bsky.social) @thserra]

Come see us at the virtual poster session of the @sparsenn workshop at #iclr2023, starting 11 AM ET

Our work is on the effect of sparsification in the number of linear regions of neural networks, and how understanding this effect may help pruning with better accuracy (Poster 30)

[Vithu Thangarasa @vithursant19]

@JeffDean discussed the importance of sparse computation, adaptive computation and dynamically-changing neural networks @sparsenn. He thinks "dense models are going to give way to these efficient sparse models"...I agree 💯

[sparsenn @sparsenn]

The next talk has just started: “Structured Sparsity Research” by Pavlo Molchanov

[sparsenn @sparsenn]

We have the next round of talks! The first talk was given by Ce Zhang #ICLR2023

[sparsenn @sparsenn]

Excited to start the next part of the workshop! We have now the breakout sessions where the attendees discuss and brainstorm various topics in sparsity with each other!
#ICLR2023

[sparsenn @sparsenn]

Our final spotlight “Automatic Noise Filtering with Dynamic Sparse Training in Deep Reinforcement Learning” is presented by @BramGrooten
#ICLR2023

[sparsenn @sparsenn]

The next spotlight presentation is “Efficient Backpropagation for Sparse Training with Speedup” presented by Mahdi Nikdan #ICLR2023

[Vithu Thangarasa @vithursant19]

I'll be presenting our paper at the #ICLR2023 Sparsity workshop @sparsenn with my co-authors @gupta__abhay and Shreyas Saxena! Great to see so many experts in the #sparsity field come together to share insights and knowledge. Come by if you're around. https://t.co/0sSvzxF6YB

[sparsenn @sparsenn]

We will have the second round of the spotlight presentations now! Starting with “Massive Language Models Can be Accurately Pruned in One-Shot” presented by Elias Franter.
#ICLR2023

[sparsenn @sparsenn]

The second session has just started with the talk by Aakanksha Chowdhery, “Efficiently Scaling Transformer Inference”.
#ICLR2023

[Anand Subramoney @anandsubramoney]

I'm presenting a poster in the Sparse Neural Network workshop @sparsenn at #ICLR2023 on "Efficient Real Time Recurrent Learning through combined activity and parameter sparsity". Come by if you're around!

Link to paper: https://t.co/ht0VYTdsQw

[sparsenn @sparsenn]

Our second spotlight presented by @ShiweiLiu9

Ten Lessons We Have Learned in the New ''Sparseland'': A Short Handbook for Sparse Neural Network Researchers

[Sophie M. Fosson @sophiefosson]

Today, I'm virtually at #ICLR2023 at SNN workshop @sparsenn to "play the lottery with concave regularizers" 🌈 #sparsity #pruning #LotteryTicketHypothesis

[sparsenn @sparsenn]

Our first spotlight happening now!

PopSparse: Accelerated block sparse matrix multiplication on IPU

Presented by Zhiyi Li
@graphcoreai

[Rosanne Liu @savvyRL]

Another poster novelty from @sparsenn workshop: "tear-off papers"! Printed, stapled, and ready to grab.

[sparsenn @sparsenn]

Panel on Understanding Sparsity with @yanii, @gkdziugait, @white_martha and @DAlistarh moderated by @Elena__Mocanu and Aleksandra Nowak.