Vineeth Gutta

Minimalist

[UD Engineering @UDengineering]

@sunitachandra29 proposes creating an in-house generative AI tool to provide access, security of data, accelerate learning & discovery and take the lead in advancing AI tech.

[Comp & Info Sciences at Univ of Delaware @UD_CIS]

Congratulations to Dr Sunita Chandrasekaran, @sunitachandra29, for being selected to serve on the U.S. Department of Energy (DOE, @ENERGY) Office of Science Advanced Scientific Computing Advisory Committee (ASCAC).

[Vineeth Gutta @vineethgutta]

I'm grateful to my co-authors from the Frederick National Lab for Cancer Research (FNLCR/NIH) and my advisor @sunitachandra29 for the invaluable guidance, insights, and support during the study.

3/3

[Vineeth Gutta @vineethgutta]

I'm thrilled to announce that my latest research paper UNNT: A novel Utility for comparing Neural Net and Tree-based models has been published in the journal PLOS Computational Biology.

Read full paper: https://t.co/t8mSUZAmbw

1/3

[Vineeth Gutta @vineethgutta]

This work presents a software for comparing Neural Net and Tree-based models enabling users with structured (tabular) data to compare CNN and XGBoost models by providing a utility to train both models using cancer drug response models as the case study.

2/3

[Tom Morgan @tomowenmorgan]

It took me 3 months to put 10 years of research on curiosity into a 20min presentation...

...So when @GrahamDuncanNYC challenged me to compress it down to 5min for the @SohnConf yesterday, I obviously went insane.

Here it is.

[Vineeth Gutta @vineethgutta]

@adrianjhpc @sunitachandra29 correct

[Vineeth Gutta @vineethgutta]

@adrianjhpc @sunitachandra29 current implementation is layer wise distribution across GPUs

[Vineeth Gutta @vineethgutta]

@adrianjhpc @sunitachandra29 model parallel

[Vineeth Gutta @vineethgutta]

@adrianjhpc @sunitachandra29 yes, that's correct

[Vineeth Gutta @vineethgutta]

@adrianjhpc @sunitachandra29 Won't have different weights on each worker. It will be a single model running across multiple Nodes. Trying to interface with another distributed HPC code that generates data on each node so trying to see if there's a way to keep the data in the node for training

[Vineeth Gutta @vineethgutta]

@NoahhPersonal Congrats!

[Richard Socher @RichardSocher]

It is kind of shocking how much current computing architectures (graphics processing units) define neural network architectures. GPUs made models like Transformers fast and hence popular, not neuroscience or theory. Our model space is restricted to what's fast on GPUs.

[Vineeth Gutta @vineethgutta]

@NoahhPersonal Congrats Noah!

[Exascale Computing Project @exascaleproject]

As principal investigator of the ECP SOLLVE project, Sunita Chandrasekaran emphasizes the development of sustainable software #exascale #supercomputing #HPC

➡️ https://t.co/UHo43s3G4o

[HPC Guru @HPC_Guru]

Congratulations @sunitachandra29!

@exascaleproject Selects Sunita Chandrasekaran @brookhavenlab as Principal Investigator of the #SOLLVE Project

https://t.co/HP4FB2LBQM

#HPC #AI #Exascale #OpenMP #GPU via @HPCwire

[V @vsoch]

I am filled with immense happiness at this discovery! For years I wanted someone to provide an alternative to Singularity Hub, and one that is free and able to support the size and quantity of containers that researchers are really building. #GitHub, you stepped up to the plate!

[clem 🤗 @ClementDelangue]

So cool to see the @deep_chem team on https://t.co/odSOQsyE2G @BrandonReeves08 @zavaindar! Any other cool bio / chem orgs or models that should be on @huggingface?

[Microsoft Research @MSFTResearch]

ZeRO-Infinity, the newest addition to DeepSpeed optimization library, supports model training w/ tens of trillions of parameters—an order-of-magnitude larger than state of the art. Learn how ZeRO-Infinity also makes training even easier w/ fewer resources: https://t.co/PAqULwoAnp

[CRPL_HPC @CRPL_HPC]

We are so proud! It could not have come to a more worthy person 🙌🥳
We are so lucky to have you as our mentor 😊 & you are a source of inspiration to all of us 💪🎉🎉