Marius Slavescu

STEMCA Inventor co-author, robotics/STEM education and DIY/maker platform, Founder of GTA Robotics community and Open Source Self Driving Car project

[Marius Slavescu @GTARobotics]

@AndrewYNg We can do it by enabling everyone to collect, analyze, share data and code! I try this in #OpenSource #SelfDrivingCar Initiative https://t.co/EUzgBBlXVg

[Andrew Ng @AndrewYNg]

For AI to be free we need not just Open Source, but also a strong Open Data movement.

[Bernt Bornich @BerntBornich]

The ability to try any new thing with a sensible approach is a cornerstone of artificial general intelligence.

Now the real world is yours for learning, NEO.

[Andre S @andreslavescu]

https://t.co/g4kcaj6mgO performance oriented changes to enable larger expansion rates (n):

1. Template specialization for n=4/8/16/32. Kernels now dispatch to optimal unroll # instead of always using 32
2. Tensor Core path for n >= 32 via cuBLASLt - restructured to row-major layout so no transpose kernels needed
3. enabling pipelining for larger expansion rates (n) for the static H path. pipelining active at n >= 16.

[Andre S @andreslavescu]

Wrote a fused mHC layer in pytorch using the kernels as a backend, and achieved some interesting speedups!

Tested on H100 SXM5

[Andre S @andreslavescu]

Just added the backward kernels

https://t.co/tdOVz8Ma6T

[Marius Slavescu @GTARobotics]

@andre_slav03 Impressive work @andre_slav03!
Implementing @deepseek_ai new revolutionary optimization approach!

[Andre S @andreslavescu]

Decided to implement Deepseek's new paper (mHC: Manifold-Constrained Hyper-Connections) in CUDA.

[Marius Slavescu @GTARobotics]

@awnihannun @awnihannun @andre_slav03 check this video, he was able to run the model on a 512GB mac + 128GB mac using:
Inferencer App: https://t.co/CIciKMgp8Q Kimi-K2-Thinking-Q4.25
https://t.co/w5JxWpYqdu

[Andre S @andreslavescu]

Update:

INT8 path now has fused perturbed matmul for better performance. MLP expand uses pure INT8 activations with no float conversion:
https://t.co/LdTZeeLyip

per-step time on H100 PCI-e is at <60ms.

@j_foerst @bidiptas13 @d0rc

[Andre S @andreslavescu]

EGGROLL 8xB200 distributed training with P2P weight averaging!

https://t.co/WimvgcOLj4

Thank you @NorthflankWill and @northflank for the B200s.

[Siddharth Dhulipalla @siddhulipalla]

It's @ArtificialAnlys official: Parakeet TDT running on @TeamHathora has the best cost to WER (word error rate) ratio of any ASR model! This was all in less than a month since we launched

https://t.co/VIaBYiY5Lm

Huge thanks to the @nvidia voice research team (Adi & folks) and the AA team (@grmcameron @kiriillbutler)

Try it for yourself here: https://t.co/5cYX6lxu4s

[Team Hathora @TeamHathora]

Tonight’s AI Lightning Talks at the @shinecapital office with talks from
Zack @AssemblyAI
@andre_slav03 @TeamHathora
And Kent from @Medal_TV and General Intuition.

We lit up the minds of NYC AI curious.

Come back in 2026!

[Siddharth Dhulipalla @siddhulipalla]

Super cool viz of how an Omni modal model can process audio tokens in/out by @andre_slav03

https://t.co/WodCSo1GGT

[Marius Slavescu @GTARobotics]

@awnihannun @awnihannun @andre_slav03 check this video, he was able to run the model on a 512GB mac + 128GB mac using:
Inferencer App: https://t.co/CIciKMgp8Q Kimi-K2-Thinking-Q4.25
https://t.co/w5JxWpYqdu

[Ethan Mollick @emollick]

Here are our three papers on ways to use AI for teaching. All have prompts & details:
Using AI to make teaching easier & more impactful: https://t.co/SZ0zBa6ZPI
Using errors made by AI to teach hard things: https://t.co/cwrNvUjTAd
Assigning AI to students: https://t.co/Un1yfuHcX3

[Indy Autonomous Challenge @IndyAChallenge]

Thank you @Autodromo_Monza for the completion certificate of the first race on a non-oval circuit during MIMO 2023 🏁🤖🏎️
#MilanoMonza #IndyAutonomousChallenge #NoSteeringWheelNoProblem
#AutonomousRacing #MIMO2023

[Indy Autonomous Challenge @IndyAChallenge]

“More than 10,000 spectators filled the stands to witness the driverless racecars competing on the Monza “Temple of Speed” Formula One circuit as part of the Milan Monza Motor Show (MIMO) that ran last weekend at the @Autodromo_Monza” - @IAutonomousV

https://t.co/qVR9hBmzxQ

[Indiana Economic Development Corp. @Indiana_EDC]

We’re showing @AndrettiFE’s @JakeDennis19 around Indy today. During our first stop, he’s seeing the innovations in autonomous mobility happening right here in Indiana with @IndyAChallenge.

[University of Waterloo @UWaterloo]

Congratulations to autonomous racing team MITT-PITT-RW who achieved a personal best time of 173.8 kph on their first @IndyAChallenge road race in Monza, Italy 🤖🏎️

The first challenge? Learning to turn right!

Full Story: https://t.co/ZW2d3hXSp3 | #UWaterlooNews

[Davide Scaramuzza @davsca1]

Congratulations to the winners and finalists of the Best Paper Award of
the 4th Workshop on Event-based Vision at #CVPR2023: Papers and code: https://t.co/HyMercJQRZ #eventcamera With @guillermogb @KostasPenn @CFermuller @DMigliorePro

[Andrej Karpathy @karpathy]

Carve out a few hours to learn https://t.co/6szj33iRBz
Powerful for rapid prototyping, interactive visualization.
It's a hammer and you'll start seeing a lot of nails.