Max Shad

Senior Director of AI/ML Research Engineering, Kempner Institute @KempnerInst @Harvard , views are my own

[Max Shad @maxshadx]

Real-time GPU monitoring made simple!

#GPU #AI

[Naeem Khoshnevis @NaeemKhoshnevis]

https://t.co/LAROtyD9z3

[Max Shad @maxshadx]

Really proud to see this announced. I led the design of the Kempner Institute's upgraded AI cluster to balance large-scale performance with the flexibility researchers need for fast-moving, day-to-day work.

With 1,144 data center GPUs and expected performance of 1.79 exaFLOPS, this is a major step forward for academic AI infrastructure. I'm excited to see how the Kempner community at Harvard University uses this new capacity to push forward research in AI, NeuroAI, robotics, and beyond.

#KempnerInstitute #Harvard #AI #NeuroAI #ML #GPU #Computing #AICluster #Compute #MachineLearning #InfiniBand #Networking

[Kempner Institute at Harvard University @KempnerInst]

🚀Kempner Institute's AI cluster gets a big upgrade! We're adding 424 H200s & 192 RTX PRO 6000 Blackwell GPUs, creating a 1,144-GPU supercomputer delivering 1.79 exaFLOPS for AI, neuroscience & robotics research. Read more: https://t.co/UI46cOHlRq
@ShamKakade6 @blsabatini @MGHPCC

[Kempner Institute at Harvard University @KempnerInst]

💡⚙️#KempnerInstitute is at #SC25 in St. Louis! Stop by the @MGHPCC booth to learn more about our work and meet members of the Kempner R&E team.
@supercomputing @MaxShadX

[Kempner Institute at Harvard University @KempnerInst]

🧠💻Kempner researchers are using the new #DGXSpark to map how genes shape the brain — bringing the power of #AI straight to the bench and accelerating the pace of discovery.🚀
https://t.co/TYIrgMMQaI
#Neuroscience @blsabatini @harvardmed @stevens1lab  @BostonChildrens

[Naeem Khoshnevis @NaeemKhoshnevis]

Proud and grateful to be contributing to such an inspiring project.

[Zahra Kadkhodaie @ZKadkhodaie]

Diffusion models learn probability densities by estimating the score with a neural network trained to denoise. What kind of representation arises within these networks, and how does this relate to the learned density? @EeroSimoncelli @StephaneMallat and I explored this question.

[Yilun Du @du_yilun]

Excited to share Equilibrium Matching (EqM)!

EqM simplifies and outperforms flow matching, enabling strong generative performance of FID 1.96 on ImageNet 256x256.

EqM learns a single static EBM landscape for generation, enabling a simple gradient-based generation procedure.

[Kempner Institute at Harvard University @KempnerInst]

New in the #DeeperLearningBlog: Kempner researchers Nikhil Anand (@nikhil_anand91) and Chloe Su (@Huangyu58589918) discuss new work on how numerical precision can impact the accuracy and stability of #LLMs.

https://t.co/SS3CDlUKYf

#AI

(1/2)

[Max Shad @maxshadx]

Learn distributed training from our GPU engineering expert @y_mazloumi and get the opportunity to earn an NVIDIA DLI certification!

@KempnerInst
@nvidia
#GPU

[Kempner Institute at Harvard University @KempnerInst]

Registration is now open for our in-person data parallelism workshop on June 4 to kick off our Frontiers in NeuroAI symposium! Led by @y_mazloumi, this workshop is an opportunity to earn your @nvidia
DLI certificate. Learn more: https://t.co/2QmPQoMm3y #GPUs #ML @maxshadx

[Max Shad @maxshadx]

Great opportunity - apply if you're interested or share with your network!

@KempnerInst

#Harvard #Kempner #AI

[Kempner Institute at Harvard University @KempnerInst]

We're hiring a Senior ML Research Engineer! We're looking for someone with with deep technical skills in ML, deep learning, and software engineering best practices. This role would be a great fit for someone who thrives in a collaborative environment.

https://t.co/23a3wF5s7o

[Max Shad @maxshadx]

A lot to learn from Kempner Institute’s Computing Handbook. Check out the materials on distributed #GPU computing.

[Kempner Institute at Harvard University @KempnerInst]

Optimize your use of high-performance #computing with the #KempnerInstitute Computing Handbook, a publicly-available online resource that offers expert guidance for honing your skills in #AI, #ML and high-performance computing. Read more: https://t.co/uGW9dcTUJo
@maxshadx

[Kempner Institute at Harvard University @KempnerInst]

Congrats to #KempnerInstitute's @marinkazitnik & colleagues whose work is featured as a spotlight paper at #NeurIPS2024! Check out "Generalized Protein Pocket Generation with Prior-Informed Flow Matching" at the 11a session today. @ZaixiZhang @leuchine https://t.co/1h1LP9vlnN

[Max Shad @maxshadx]

Eleven years ago, I named our first HPC cluster "Ludwig" to honor his contributions to science. With its computing power, we simulated complex fluid dynamics using Lattice Boltzmann Method!

#science #hpc #fluid

[Physics In History @PhysInHistory]

Ludwig Boltzmann’s grave in Vienna bears his famous entropy formula, S = k log W, which relates the entropy of a system to the number of possible microscopic states.
Ludwig Boltzmann was an Austrian physicist and philosopher who made significant contributions to the fields of statistical mechanics and thermodynamics including the formulation of the H-theorem- which explains the behavior of gases in terms of statistical mechanics, the concept of the Boltzmann constant (k)- a fundamental constant in physics that relates temperature to the average kinetic energy of particles in a gas, the Boltzmann equation- which describes the behavior of a gas by incorporating statistical probabilities.
The concept of a "Boltzmann Brain" was introduced by cosmologists, inspired by Boltzmann's work. It suggests that there is a minuscule possibility of a conscious entity, like a human brain, spontaneously fluctuating into existence in an otherwise empty and chaotic universe. This idea sparked debates and discussions about the nature of reality and the interpretation of probabilities in cosmology.

[Max Shad @maxshadx]

Compute power supporting @KempnerInst research! Now on TOP500 and Green500 list. #SC24

[Kempner Institute at Harvard University @KempnerInst]

We're proud to share the #KempnerInstitute’s AI cluster has joined the ranks of the fastest green supercomputers in the world (32nd on Green500 and 85th in TOP500)! Read the full announcement: https://t.co/0eNgnwtmkm #ML #computing #SC24

[Kempner Institute at Harvard University @KempnerInst]

We're here at @Supercomputing this week! Stop by the @MGHPCC booth to learn about our AI cluster and say hello to some of the #KempnerInstitute's research & engineering team. @y_mazloumi @maxshadx #SC24

[Kempner Institute at Harvard University @KempnerInst]

Attending @Supercomputing this week in Atlanta? Be sure to stop by the @MGHPCC booth to learn about the #KempnerInstitute AI cluster, and more about our educational programs & other opportunities. Hope to see you then! @maxshadx @y_mazloumi
https://t.co/yMfik3KVUg

[Max Shad @maxshadx]

I will be at #SC24 next week in Atlanta. Happy to chat about #supercomputing in #AI if you’re attending!

[SC26 @Supercomputing]

Join us for an unparalleled experience in the HPC community! Attend technical presentations, workshops, tutorials, posters, and Birds of a Feather sessions. Discover cutting-edge innovations from top manufacturers, research organisations, and universities. https://t.co/OpD2pJfKAC

[Kempner Institute at Harvard University @KempnerInst]

We're hiring! Our Associate Director of Educational Programs oversees all aspects of the #KempnerInstitute's fellowship and training programs for undergraduate, post-baccalaureate, and graduate students. Apply today: https://t.co/lEbVsAdpWm #AI #education @Harvard

[Max Shad @maxshadx]

Attention is all they need!

#ginger #mango

[Sham Kakade @ShamKakade6]

1/5⚡Introducing Flash Inference: an *exact* method cutting inference time for Long Convolution Sequence Models (LCSMs) to near-linear O(L log² L) complexity! Faster inference, same precision. Learn how we accelerate LCSM inference.