Arash Vahdat @ArashVahdat Twitter profile | Pikagi

Pikagi

Arash Vahdat

@ArashVahdat

8,307

Followers

808

Following

118

Media

1,403

Statuses

Principal scientist and research manager @nvidia research, leading forward-looking fundamental generative AI research efforts, views are my own.

San Francisco Bay Area

https://t.co/D4aoPxWMw5

Joined January 2010

Don't wanna be here? Send us removal request.

Pinned Tweet

@ArashVahdat

Arash Vahdat

3 months

📢🔥 My team at NVIDIA Research is looking for full-time research scientists & summer interns. Topics of interest are: 1⃣Gen AI for science (climate, biology, chemistry) 2⃣Image/Video/3D (generate, edit, manipulate) 3⃣Fundamental generative learning Apply via links below

Tweet media one

11

84

553

Last Seen Profiles

@LosBrolin

@_charliehagan

@Blmanshack

@krriderbaseball

@ChrisGskyway

@hathaway0616

@pepeonerc

@outkiel

@Evan_Schiller

@calebchisum

@ApachesACP

@misscatarro

@asmtasha

@trishgreenhalgh

@CrookedMedia

@SabalenkaA

@ooc_big

@SabalenkaA

@CCS__Baseball

@TitmouseInc

@GFCSBSB

@sabicsfs

@blb

@CrowdCompass

@the_halle

@vishvagaurav

@ishida_atsuko

@stw_pdg

@mr_absolutblank

@SchwiftyStudios

@tenomnft

@ashita_utatane

@BecklnReaILife

@JohannKurtz

@FindingNovyon

@davelackie

@ArashVahdat

Arash Vahdat

3 years

I can't believe after this many years of programming with NumPy/PyTorch/TensorFlow, I didn't know about 𝚎𝚒𝚗𝚜𝚞𝚖. You can get rid of so many lines of reshape, transpose, sum, product & expand_dim with a single einsum which is even easier to understand:

Tweet media one

27

279

2K

@ArashVahdat

Arash Vahdat

4 years

📢📢📢 Introducing NVAE 📢📢📢 We show that deep hierarchical VAEs w/ carefully designed network architecture, generate high-quality images & achieve SOTA likelihood, even when trained w/ original VAE loss. paper: with @jankautz at @NVIDIAAI (1/n)

Tweet media one

15

301

1K

@ArashVahdat

Arash Vahdat

7 months

Our #CVPR2023 tutorial on diffusion models is now available publicly (with much better audio quality 🤣) Website: With: @chenlin_meng and @baaadas @CVPR

Tweet card media

Denoising Diffusion-based Generative Modeling: Foundations and Applications

Tutorial on Diffusion Models in Conjunction with CVPR 2023

cvpr2023-tutorial-diffusion-models.github.io

3

155

677

@ArashVahdat

Arash Vahdat

4 years

It breaks my 💚 when researchers tell me that VAEs don't work. My first typical question is "did you try hierarchial VAE or vanilla VAE?", the answer is usually vanilla VAE. VAEs work much better with hierarchical structures. NVAEs and this work take this to the extreme!

@hardmaru

hardmaru

4 years

Very Deep VAEs Generalize Autoregressive Models and Can Outperform Them on Images “Very Deep VAEs” achieve higher likelihoods, use fewer parameters, generate samples 1000x faster, and are more easily applied to hi-res images, compared to PixelCNN.

Tweet media one

Tweet media two

Tweet media three

5

125

550

8

92

590

@ArashVahdat

Arash Vahdat

2 years

📢📢 The recording of our tutorial on denoising diffusion models is now available on YouTube: Slides and additional info: This tutorial was originally presented by @RuiqiGao , @karsten_kreis , and myself at #CVPR2022 .

Tweet media one

6

110

532

@ArashVahdat

Arash Vahdat

2 years

🎉 Our tutorial proposal on 𝗗𝗲𝗻𝗼𝗶𝘀𝗶𝗻𝗴 𝗗𝗶𝗳𝗳𝘂𝘀𝗶𝗼𝗻-𝗯𝗮𝘀𝗲𝗱 𝗚𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗠𝗼𝗱𝗲𝗹𝗶𝗻𝗴 has been accepted to #CVPR2022 . Stay tuned for some cool lectures from @karsten_kreis , @RuiqiGao and me at @CVPR 2022.

Tweet media one

3

73

513

@ArashVahdat

Arash Vahdat

5 years

This is me while explaining relaxed Boltzmann machines and not knowing that the pioneer who introduced them to the field is just behind me :( #NeurIPS2018

@_willfalcon

William Falcon ⚡️

5 years

When you explain your poster and don’t notice Geoffrey Hinton standing behind you #NeurIPS2018

Tweet media one

24

317

2K

7

38

460

@ArashVahdat

Arash Vahdat

6 months

📢 Would you like to build latent diffusion models for new problem domains? Wondering about theoretical and practical considerations! @RuiqiGao , @karsten_kreis and I will present the #NeurIPS2023 tutorial on "Latent Diffusion Models" on Monday, Dec 11.

Tweet card media

Latent Diffusion Models: Is the Generative AI Revolution Happening in Latent Space?

Tutorial in Conjunction with NeurIPS 2023

neurips2023-ldm-tutorial.github.io

3

62

427

@ArashVahdat

Arash Vahdat

7 months

📢🔥 My team at NVIDIA research is looking for candidates with a fundamental generative learning background (ideally) in one of these domains: - Gen AI for science (climate, chemistry, biology) - Gen AI for 3D data Apply via:

Tweet media one

17

68

427

@ArashVahdat

Arash Vahdat

5 months

📢🔥My team at NVIDIA Research is looking for Summer 2024 interns. Topics of interest are: 1⃣ Fundamental generative learning (diffusion, etc) 2⃣Gen AI for science (climate, biology, chemistry) 3⃣Image/Video/3D (generate, edit, manipulate) Apply via

Tweet media one

3

62

418

@ArashVahdat

Arash Vahdat

4 years

📢 NVAE source code is released! NVAE is deep hierarchical VAE w/ specially designed network architecture that can generate high-quality images & achieve SOTA log-likelihood. Happy coding! paper: code: w/ @jankautz at @NVIDIAAI

Tweet media one

3

94

416

@ArashVahdat

Arash Vahdat

10 months

📢 I am looking for Fall interns to join the fundamental gen AI research team at NVResearch. Topics of interest are: -Fundamental Gen AI research -3D generation -Text-based manipulation Remote/in-person, full/part-time options are available. Apply at

Tweet media one

11

77

377

@ArashVahdat

Arash Vahdat

2 years

Recent text-to-image models such as #dalle2 or #Imagen are trained with ~800m image-text pairs. If that data was a movie (at 10 fps), it would take us 2.5 years to watch it all non-stop.

13

29

375

@ArashVahdat

Arash Vahdat

4 years

NVIDIA is bringing GANtastic tools for video conferencing for you!

5

60

349

@ArashVahdat

Arash Vahdat

3 years

📢 Latent Score-based Generative Model (LSGM) Score-based generative models (SGMs) are applied directly in data space & often require 1000s of network evaluations for sampling. We introduce LSGM for training SGMs in a latent space. w/ @karsten_kreis (eq. contrib.) & @jankautz

Tweet media one

8

67

309

@ArashVahdat

Arash Vahdat

2 years

🔥Denoising Diffusion GAN🔥 We tackle the 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝘁𝗿𝗶𝗹𝗲𝗺𝗺𝗮 with the novel 𝗱𝗲𝗻𝗼𝗶𝘀𝗶𝗻𝗴 𝗱𝗶𝗳𝗳𝘂𝘀𝗶𝗼𝗻 𝗚𝗔𝗡, a diffusion model designed for fast sampling. Zhisheng Xiao @karsten_kreis ft. Peanut 🐈

5

60

299

@ArashVahdat

Arash Vahdat

2 years

Check out our brief introduction to 𝗱𝗶𝗳𝗳𝘂𝘀𝗶𝗼𝗻 𝗺𝗼𝗱𝗲𝗹𝘀 and the summary of the cutting-edge research we do in this space at @nvidia .

Tweet card media

Improving Diffusion Models as an Alternative To GANs, Part 1 | NVIDIA Technical Blog

Part 1 of this series introduces diffusion models as a powerful class for deep generative models and examines their trade-offs in addressing the generative learning trilemma.

developer.nvidia.com

5

62

293

@ArashVahdat

Arash Vahdat

2 years

📢 #LACE : Controllable & Compositional Generation with Latent-Space Energy-Based Models We introduce an extremely simple method for converting unconditional GANs into conditional models by training only a classifier. abs: project:

3

53

292

@ArashVahdat

Arash Vahdat

3 years

My favorite #NeurIPS2020 tutorial is "Deep Implicit Layers" from @zicokolter , @DavidDuvenaud & @SingularMattrix . The quality of the presentation, supplementary notes, and example codes sets the bar so high. Well done! Website:

Deep Implicit Layers - Neural ODEs, Deep Equilibirum Models, and Beyond

This web page is the companion website to our NeurIPS 2020 tutorial, created by [Zico Kolter](http://zicokolter.com), [David Duvenaud](http://www.cs.toronto.edu/~duvenaud/), and [Matt Johnson](http...

implicit-layers-tutorial.org

1

46

277

@ArashVahdat

Arash Vahdat

4 years

🔥 VAEBM = VAE + EBM 🔥 VAEBM is a new generative model defined by a symbiotic composition of VAE and EBM. VAEBM improves generative quality of VAEs & explicit EBMs by a large margin & reduces the gap with GANs. w/ Zhisheng Xiao, Karsten Kreis, @jankautz

Tweet media one

2

73

277

@ArashVahdat

Arash Vahdat

9 months

At #ICML2023 , I'm hiring Fall interns & research scientists for fundamental gen AI roles. If your background is relevant, shoot me an email with your CV. Job postings:

8

30

256

@ArashVahdat

Arash Vahdat

2 years

📢 Diffusion Models for Adversarial Purification ( #ICML2022 ) Given an adversarial image, our DiffPure removes adv. perturbations by adding noise following the forward diffusion process, then denoising using the reverse generative process. project page:

Tweet media one

4

45

247

@ArashVahdat

Arash Vahdat

4 years

NVAE's accepted to #NeurIPS2020 as a #spotlight paper! Many thanks to anonymous reviewers & AC who recognized the potential impact of NVAE The code is already released here: You'll soon hear about exciting stuff we've been doing. w/ @jankautz @NVIDIAAI

Tweet card media

GitHub - NVlabs/NVAE: The Official PyTorch Implementation of "NVAE: A Deep Hierarchical Variational...

The Official PyTorch Implementation of "NVAE: A Deep Hierarchical Variational Autoencoder" (NeurIPS 2020 spotlight paper) - NVlabs/NVAE

@ArashVahdat

Arash Vahdat

4 years

📢📢📢 Introducing NVAE 📢📢📢 We show that deep hierarchical VAEs w/ carefully designed network architecture, generate high-quality images & achieve SOTA likelihood, even when trained w/ original VAE loss. paper: with @jankautz at @NVIDIAAI (1/n)

Tweet media one

15

301

1K

6

35

240

@ArashVahdat

Arash Vahdat

5 months

🔥📢 our #NeurIPS2023 tutorial on Latent Diffusion Models will take place tomorrow at 9:45 am at Hall E2. Hope to see you all there! w/ @karsten_kreis & @RuiqiGao

Tweet card media

Latent Diffusion Models: Is the Generative AI Revolution Happening in Latent Space?

Tutorial in Conjunction with NeurIPS 2023

neurips2023-ldm-tutorial.github.io

2

39

238

@ArashVahdat

Arash Vahdat

4 years

📢📢 NCP-VAE: VAEs w/ Noise Contrastive Priors We tackle the prior hole problem: prior's failure to match the aggregate posterior. We train a reweighting term in prior using noise contrastive estimation. w/ @JyotiAneja @alexschwing @jankautz

Tweet media one

7

54

205

@ArashVahdat

Arash Vahdat

3 years

@ilyasut until receiving the first AWS bill 😂

3

1

192

@ArashVahdat

Arash Vahdat

3 years

Likelihood-based generative learning is really really hard! Covering all the data distribution while having high sample quality and fast generation is probably one of the biggest challenges that the research community is facing currently.

4

15

174

@ArashVahdat

Arash Vahdat

2 years

Are you working on score-based models for: 1⃣Generative diffusion models 2⃣Differential eqn 3⃣Optimal transport 4⃣Stein’s method 5⃣Cool applications Consider submitting your work to our #NeurIPS2022 workshop on score-based methods: Deadline: 22 Sept 2022

Tweet media one

0

23

156

@ArashVahdat

Arash Vahdat

11 months

📢 #CVPR2023 Spontaneous Diffusion Meetup Several folks have asked me to do a casual meetup for those interested in diffusion models @CVPR . Let's meet tomorrow (Wed) at 1 pm in West building (upper levels). Exact location will be posted below this tweet. Please spread the word

10

33

152

@ArashVahdat

Arash Vahdat

1 month

Jensen's energy and passion are contagious! I don't know how he can always lift everyone around him up.

Tweet media one

1

2

151

@ArashVahdat

Arash Vahdat

11 months

📢 We are looking for candidates with a strong background in generative learning on text, image, video, graph & 3D data to join our team @nvidia . We strive for high-impact forward-looking research. Apply via: Please share!

Tweet media one

4

43

148

@ArashVahdat

Arash Vahdat

3 years

The last 4-5 weeks have been stressful for me since my dad in his late 70s was diagnosed with #COVID19 , and my mom, also in her 70s, refused to live separately away from my dad while he was going through this. My parents live in Iran, ~12,000 KMs away from where I am.

17

4

149

@ArashVahdat

Arash Vahdat

2 years

📢 Our team is hiring research scientists in various areas in machine learning & computer vision including 𝗮𝗱𝘃𝗮𝗻𝗰𝗲𝗱 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗺𝗼𝗱𝗲𝗹𝘀. Please apply below especially if you have worked on generative models such as diffusion models.

2

30

146

@ArashVahdat

Arash Vahdat

2 years

Discrete math is one of the reasons I hated computer engineering in undergrad. I was determined to change my major to mechanical or electrical engineering because of their core math training. I changed my mind later after taking computer vision & signal processing courses.

@CSProfKGD

Kosta Derpanis

2 years

#KostasThoughts : Core math in many #ComputerScience programs is inadequate for studying #ComputerVision & #MachineLearning . Want to study these subjects? Take (single/multi-variable) calculus, linear algebra, statistics & probability. Your future self will thank you 😉

15

27

214

7

10

130

@ArashVahdat

Arash Vahdat

1 year

Check our #GTC23 session on "GANs to Diffusion - the path to #generativeAI " where Mitesh Patel and I talk about gen AI's past, present, and future. We discuss deep generative models, their shortcomings, and how diffusion models helped us solve them. link:

Tweet media one

1

21

134

@ArashVahdat

Arash Vahdat

4 years

This was really fast! In less than 15 hours after releasing the paper, @ykilcher made a video on NVAE.

@ykilcher

Yannic Kilcher 🇸🇨

4 years

Were you always suspicious of VAEs? 👀 Too blurry? Unstable? 😱 Well stop right there, because @NVIDIAAI has built NVAE, a hierarchical multi-scale VAE that can output crispy clear samples at high resolution 💃🔥 Watch the Video! @ArashVahdat @jankautz

Tweet media one

7

45

277

3

14

130

@ArashVahdat

Arash Vahdat

3 years

🔥 #ICLR2021 #Spotlight 🔥 Our VAEBM has been accepted to ICLR as a spotlight. VAEBM was results of an amazing collaboration with Zhisheng Xiao (University of Chicago), Karsten Kreis (NVIDIA's Toronto AI lab) & @jankautz (NVResearch Boston) spanning 4 locations & 3 timezones

@ArashVahdat

Arash Vahdat

4 years

🔥 VAEBM = VAE + EBM 🔥 VAEBM is a new generative model defined by a symbiotic composition of VAE and EBM. VAEBM improves generative quality of VAEs & explicit EBMs by a large margin & reduces the gap with GANs. w/ Zhisheng Xiao, Karsten Kreis, @jankautz

Tweet media one

2

73

277

2

20

131

@ArashVahdat

Arash Vahdat

5 years

I am thrilled to announce that I have joined @nvidia research as a senior research scientist. I will continue tackling challenging problems in machine learning and computer vision along with brilliant @NvidiaAI researchers @jankautz @AnimaAnandkumar @liu_mingyu

16

4

121

@ArashVahdat

Arash Vahdat

5 years

looking for an intern to work with me at NVIDIA research next summer on one of these topics: - likelihood-based generative learning - Probabilistic weakly-supervised models (e.g., noisy labels) - Representation learning - Discrete latent models - Neural Architecture Search 👇

1

26

119

@ArashVahdat

Arash Vahdat

4 years

📢📢 Join us for our #ECCV2020 virtual tutorial on "New Frontiers for Learning with Limited Labels or Data" where I will talk about training with noisy labels! With @shalinidemello @SifeiL Varun Jampani @PavloMolchanov @ZhidingYu @YujiaHuangC @AnimaAnandkumar @jankautz @NVIDIAAI

@shalinidemello

Shalini De Mello

@shalinidemello

4 years

How can we learn vision tasks with noisy or no labels? What if we have only a pre-trained network and no data to learn from? Explore these and more exiting new frontiers for learning with limited labels or data in our upcoming #ECCV2020 tutorial:

Tweet media one

1

60

218

3

21

122

@ArashVahdat

Arash Vahdat

4 years

If you are a Ph.D. student, consider applying to NVIDIA's graduate fellowship program, especially if you're working on artificial intelligence, robotics, autonomous vehicles, and related fields.

0

38

115

@ArashVahdat

Arash Vahdat

3 years

@suzatweet Vancouver! Hiking and sky mountains are even accessable with public transit.

Tweet media one

2

0

114

@ArashVahdat

Arash Vahdat

5 years

What I was about to share: Yeayy! I got free registration to NeurIPS as one of the top reviewers What I wasn't going to share: 2 papers rejected. I forgot that AAAI deadline was last week Morality: Twitter shows the successful face of us. Never get fooled! Never give up!

1

5

114

@ArashVahdat

Arash Vahdat

1 year

There should be a process to penalize irresponsive reviewers. There are a couple of ICML reviewers in my batch that never bothered to submit their reviews and never replied to my emails/messages when asking about it. This happens at other conferences too.

12

2

108

@ArashVahdat

Arash Vahdat

2 months

Today, Jensen announced our work on CorrDiff in the #GTC24 keynote. CorrDiff is a new generative model that superresolves weather events from 25km to 2km resolution, with 1,000x the speed and 3,000x the energy efficiency of conventional models.

Tweet card media

NVIDIA CorrDiff: Resolving Extreme Weather Events With Generative AI

NVIDIA's Earth-2 Inference service is revolutionizing extreme weather forecasting by enabling users to combine global medium-range AI weather prediction mode...

www.youtube.com

1

16

106

@ArashVahdat

Arash Vahdat

1 year

At #NeurIPS2022 , we are holding a workshop on score-based methods and diffusion models that will feature a panel discussion among experts in this space. Help us ask them intriguing questions by filling out this form: More info at:

Tweet card media

NeurIPS 2022 Workshop on Score-Based Methods

Questions for panelists of the NeurIPS 2022 Workshop on Score-Based Methods (SBM @ NeurIPS 2022). Panelists: Rianne van den Berg, Microsoft Research Mark Chen, OpenAI Arnaud Doucet, University of...

docs.google.com

1

21

105

@ArashVahdat

Arash Vahdat

5 years

The code for "Learning Undirected Posteriors by Backpropagation through MCMC" is released. I had lots of fun working on this. The paper comes with in-depth discussion of possible future works, ideal for summer interns😉 paper code

Tweet media one

0

14

105

@ArashVahdat

Arash Vahdat

2 years

You can now check our preliminary slides for the @CVPR tutorial on #diffusion_models on our website: See you at 8:30 am CDT on Sunday at #CVPR2022 in New Orleans or on the virtual platform. w/ @karsten_kreis & @RuiqiGao

Tweet media one

3

15

98

@ArashVahdat

Arash Vahdat

5 years

Geoffrey Hinton is such a humble and humorous person! Finally, got the chance to hear his thoughts about our work on marrying Boltzmann machines and deep generative learning #NeurIPS2018

Tweet media one

2

9

94

@ArashVahdat

Arash Vahdat

4 years

If you are interested in combining energy-based models and VAEs, check out our #ICML2020 paper that proposes an approach for training EBMs as approximate posteriors: poster session: Wed 8 am and 9 pm PT.

@ArashVahdat

Arash Vahdat

4 years

Our work on training undirected graphical models (UGMs) as approximate posteriors is accepted to #ICML2020 . We show that UGMs can be trained by backpropagation through MCMC steps. We examine these ideas on RBM approximate posteriors in VAEs paper w/ code

0

9

57

0

19

94

@ArashVahdat

Arash Vahdat

2 years

Machine learning Twitter: how do you pass a large list of arguments to your python training scripts? If you are happy with any other library please comment below.

yaml

674

ml_collections

72

argparse

723

absl.flags

51

49

6

93

@ArashVahdat

Arash Vahdat

2 years

Today at #ICLR2022 , we are presenting two #spotlight papers on denoising diffusion models 👇

Tweet media one

1

18

92

@ArashVahdat

Arash Vahdat

2 years

We have released the #LSGM source code with a brand new project page at: If you're interested in denoising diffusion models (a.k.a. score-based generative models), come to our poster session next week at #NeurIPS2021 .

Tweet media one

1

15

91

@ArashVahdat

Arash Vahdat

2 years

It is really cool to see that denoising diffusion GANs have already been applied to speech synthesis just 4 months after our paper became available on OpenReview 🤯

@_akhaliq

AK

2 years

DiffGAN-TTS: High-Fidelity and Efficient Text-to-Speech with Denoising Diffusion GANs abs:

Tweet media one

0

25

125

0

9

91

@ArashVahdat

Arash Vahdat

3 years

eigops takes this to another level:

2

9

88

@ArashVahdat

Arash Vahdat

3 years

Post #ICLR2022 deadline bike ride! 🚴‍♂️

Tweet media one

2

0

89

@ArashVahdat

Arash Vahdat

1 year

📢 We are organizing the #CVPR2023 tutorial on diffusion models and need your help in selecting the content:

@baaadas

Jiaming Song

1 year

@chenlin_meng , @ArashVahdat , and I are presenting the #diffusion model tutorial at #CVPR2023 on June 18 (). Since there are > 1300 papers on this topic, we cannot read all of them😭, and we need your help on uncovering all the "hidden gems"!

8

51

224

2

9

86

@ArashVahdat

Arash Vahdat

2 years

The source code for our denoising diffusion GAN is released: code: project page: Happy coding! 🧑‍💻

Tweet card media

Tackling the Generative Learning Trilemma with Denoising Diffusion GANs

To reduce the number of sampling steps in diffusion models, we propose to model the denoising distribution with conditional GANs. We show our model tackles the generative learning trilemma & achieves...

nvlabs.github.io

@ArashVahdat

Arash Vahdat

2 years

🔥Denoising Diffusion GAN🔥 We tackle the 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝘁𝗿𝗶𝗹𝗲𝗺𝗺𝗮 with the novel 𝗱𝗲𝗻𝗼𝗶𝘀𝗶𝗻𝗴 𝗱𝗶𝗳𝗳𝘂𝘀𝗶𝗼𝗻 𝗚𝗔𝗡, a diffusion model designed for fast sampling. Zhisheng Xiao @karsten_kreis ft. Peanut 🐈

5

60

299

1

15

86

@ArashVahdat

Arash Vahdat

4 years

What?!

Tweet media one

3

8

86

@ArashVahdat

Arash Vahdat

3 years

Check out our new #cvpr2021 paper! We propose a simple technique to recover training images from gradients during training, questioning the efficacy of federated learning techniques in preserving data privacy.

@yin_hongxu

Hongxu (Danny) Yin

3 years

Happy to introduce GradInversion #CVPR2021 - gradients encode a surprisingly large amount of information, such that all the individual images can be completely recovered, even for complex ImageNet, deep networks, over large batch sizes. Paper:

1

21

90

0

12

85

@ArashVahdat

Arash Vahdat

11 months

🔥🔥 Jiaming is telling us all about diffusion models for 3D, video, and beyond! #CVPR2023 @CVPR

Tweet media one

@baaadas

Jiaming Song

11 months

🔥

Tweet media one

0

1

40

2

7

79

@ArashVahdat

Arash Vahdat

5 years

Did you know when you take the geometric mean of several distributions, you're minimizing a KL divergence? In this post, I show how the geometric mean is related to the KL divergence, and how we can use this relation for training from noisy labels. Post:

2

17

82

@ArashVahdat

Arash Vahdat

3 years

LSGM will be presented at #NeurIPS2021 🎉 In this work, we show how score-based generative models (a.k.a. denoising diffusion models) can be trained in a latent space.

@ArashVahdat

Arash Vahdat

3 years

📢 Latent Score-based Generative Model (LSGM) Score-based generative models (SGMs) are applied directly in data space & often require 1000s of network evaluations for sampling. We introduce LSGM for training SGMs in a latent space. w/ @karsten_kreis (eq. contrib.) & @jankautz

Tweet media one

8

67

309

3

4

82

@ArashVahdat

Arash Vahdat

2 years

book your ✈️, reserve your 🏨, pack your 🛄, mark your 🗓️! 📢 #diffusion_models are coming to #CVPR2022 on Sunday, June 19. Check out our @CVPR tutorial page at: w/ @RuiqiGao & @karsten_kreis

Tweet media one

1

14

79

@ArashVahdat

Arash Vahdat

7 months

Kosta sets the bar so high with the quality of his lectures!

@CSProfKGD

Kosta Derpanis

7 months

Teaser of my lecture on Diffusion Models (ver. 1.0, first complete draft) 😅

9

16

172

0

3

74

@ArashVahdat

Arash Vahdat

4 years

I have a joke about VAEs, but my memory is blurry

1

6

73

@ArashVahdat

Arash Vahdat

2 years

Impressive story! Fun fact: I got rejected from TTIC the same year 😜

@bneyshabur

Behnam Neyshabur

2 years

Totally agree! Anyone screening applications and any applicant thinking their CV is not representative of their skills/potentials, I think you might want to read the story of my own PhD application in this thread: 1/

30

171

1K

3

1

73

@ArashVahdat

Arash Vahdat

11 months

🔥🔥🔥Join our #CVPR2023 tutorial on diffusion models, happening tomorrow morning!

@baaadas

Jiaming Song

11 months

📢 Our #CVPR2023 tutorial on "Denoising Diffusion Models: A Generative Learning Big Bang" w/ @chenlin_meng and @ArashVahdat is happening tomorrow morning! 9:00 to 12:30, West 202-204. This is the year of big bang for diffusion models in CVPR!

Tweet media one

6

37

198

3

7

72

@ArashVahdat

Arash Vahdat

3 years

. @iclr_conf this year allowed self-nomination for Area Chairs which gave me a chance to nominate myself and serve as a junior AC after 10 years of publishing at top conferences. It would be great if other conferences like @NeurIPSConf , @icmlconf , @CVPR did the same.

@EmtiyazKhan

Emtiyaz Khan

3 years

I was talking to an ECR (PI level) who has reviewed and published in major ML conferences for years, but never been invited as an Area Chair. I am wondering how to improve this. Do others relate to this? Happy to hear your opinion (DM me if uncomfortable tweeting in public).

3

0

28

1

4

72

@ArashVahdat

Arash Vahdat

1 year

I hope you are having an amazing time at #NeurIPS2022 workshop on score-based methods. We are in Rooms 293 and 294.

Tweet media one

2

7

71

@ArashVahdat

Arash Vahdat

3 years

The 4-year nightmare is over!

1

0

71

@ArashVahdat

Arash Vahdat

2 years

Yesterday we ran into several connection issues & we couldn't stream our @CVPR tutorial on diffusion models properly online. I apologize for all the inconvenience. We are going to redo all our presentations and post them to the website in the coming weeks. Stay tuned! #CVPR2022

@ArashVahdat

Arash Vahdat

2 years

You can now check our preliminary slides for the @CVPR tutorial on #diffusion_models on our website: See you at 8:30 am CDT on Sunday at #CVPR2022 in New Orleans or on the virtual platform. w/ @karsten_kreis & @RuiqiGao

Tweet media one

3

15

98

1

7

71

@ArashVahdat

Arash Vahdat

3 years

The current status of election result reporting.

@lifesnoteasyman

Sour Man

@lifesnoteasyman

3 years

Who did this?? :))))))))))))))))))))))) #Nevada

38

448

3K

5

4

71

@ArashVahdat

Arash Vahdat

3 years

My road bike was stolen a few months ago, so I got a new Trek Domane. This bike is one of the popular bikes used in UNPAVED segments of Tour De France, which makes it a perfect bike for PAVED roads in the bay area 🤦‍♂️

Tweet media one

5

1

70

@ArashVahdat

Arash Vahdat

3 years

I am very happy that my family managed to beat covid, and did so without transmitting it to each other so far. Folks, take this disease seriously! I hope none of you and your loved ones go through #COVID19 .

5

2

70

@ArashVahdat

Arash Vahdat

2 years

Our score-based generative modeling with critically-damped Langevin diffusion paper will be presented at #ICLR2022 as a spotlight paper! 🎉 Amazing work with @timudk & @karsten_kreis !

@karsten_kreis

Karsten Kreis

2 years

📢 Score-based Generative Modeling with Critically-Damped Langevin Diffusion! We propose a novel diffusion using auxiliary velocity variables for more efficient denoising and higher quality generative models. w/ the amazing @timudk & @ArashVahdat ! (1/n)

7

58

212

1

6

66

@ArashVahdat

Arash Vahdat

2 years

And, the Oscar award goes to CMT that went down just before the #CVPR2022 deadline.

1

7

67

@ArashVahdat

Arash Vahdat

5 months

Peanut, the diffusion cat was featured in a nice talk by @RuiqiGao at #NeurIPS2023

Tweet media one

4

5

65

@ArashVahdat

Arash Vahdat

3 years

📢 HANT: Hardware-Aware Network Transformation How can we accelerate a trained network to meet efficiency requirements for deployment? The current network compression methods such as pruning, kernel fusion & quantization address this w/o changing underlying network operations.

Tweet media one

1

10

64

@ArashVahdat

Arash Vahdat

3 years

These visualizations are mesmerizing!

@_akhaliq

AK

3 years

Alias-Free GAN pdf: project page: networks match the FID of StyleGAN2 but differ dramatically in their internal representations, and they are fully equivariant to translation and rotation even at subpixel scales

12

325

1K

0

7

62

@ArashVahdat

Arash Vahdat

1 year

I'm attending #NeurIPS2022 in person this week. I hope to meet many of you in person finally!!!

4

2

61

@ArashVahdat

Arash Vahdat

3 years

If you missed our talk on NVAE last night at #NeurIPS2020 , you can watch the full video here:

@NVIDIAAI

NVIDIA AI

3 years

Introducing "NVAE", the first successful VAE applied to natural images as large as 256×256 pixels! @ArashVahdat @JanKautz Watch the full-length demo of NVAE here: #NeurIPS2020

0

15

63

0

6

61

@ArashVahdat

Arash Vahdat

4 years

So this week, the #ML community is attending #ICLR2020 , while discussing rebuttals for the next conference #ICML2020 , preparing their drafts for the next next conference #NeurIPS2020 and wondering how to slow down.

0

5

62

@ArashVahdat

Arash Vahdat

3 years

My tensorboard results are updated more rapidly than this election's results.

0

2

61

@ArashVahdat

Arash Vahdat

2 months

I'm attending #GTC24 for the first time in person, and I'm genuinely impressed by the preparation that the green team has put into this. Well done @nvidia !

Tweet media one

0

0

61

@ArashVahdat

Arash Vahdat

2 years

My heart is with brave Iranian people who are fighting for their most basic rights! #Woman_Life_Freedom #زن_زندگی_آزادی

@DUALIPA

DUA LIPA

2 years

Sarina Esmailzadeh, 16, beaten to death by Iranian security forces for protesting for women’s rights in Iran. Nika Shakarami, 16, killed after burning her headscarf in protest. Hadis Najafi, 23, shot multiple times during demonstrations sparked by the death of Mahsa Amini, 22,

Tweet media one

Tweet media two

Tweet media three

Tweet media four

3K

36K

124K

1

5

59

@ArashVahdat

Arash Vahdat

4 years

Happen to have experience with neural architecture search? 👉 Send me your CV if you are interested in working on NAS as a summer intern at @NvidiaAI research. 👉 Burned out with #ICML2020 and #CVPR2020 deadlines? Get back to me only when you are well-rested!!

1

18

60

@ArashVahdat

Arash Vahdat

3 years

It turned out that the #NeurIPS2021 supplementary material deadline can be more stressful than the main conference deadline when you have an ∞ number of pages in the appendices.

1

0

59

@ArashVahdat

Arash Vahdat

3 years

Confirmed! #NeurIPS2021 reviews will be done on openreview! This is great change that hopefully will encourage more discussion during the review period!

@openreviewnet

Open Review

3 years

@tetraduzione @ArashVahdat @NeurIPSConf It is true! We are very excited that we are hosting NeurIPS this year. We hope to do our best to improve the AC experience.

1

7

55

1

0

57

@ArashVahdat

Arash Vahdat

4 years

Our work on training undirected graphical models (UGMs) as approximate posteriors is accepted to #ICML2020 . We show that UGMs can be trained by backpropagation through MCMC steps. We examine these ideas on RBM approximate posteriors in VAEs paper w/ code

0

9

57

@ArashVahdat

Arash Vahdat

9 months

~5 hours to the panel discussion at the #ICML2023 Workshop on Structured Probabilistic Inference & Generative Modeling:

Tweet media one

1

0

57

@ArashVahdat

Arash Vahdat

2 years

@rasbt IMO they're a bit hard to pick up requiring working knowledge of SDEs/ODEs/VAE/norm. flows. They're very easy & stable to train, but also slow to sample from. These two ICLR papers have reduced their sampling cost dramatically:

Tweet card media

Tackling the Generative Learning Trilemma with Denoising Diffusion GANs

To reduce the number of sampling steps in diffusion models, we propose to model the denoising distribution with conditional GANs. We show our model tackles the generative learning trilemma & achieves...

nvlabs.github.io

1

6

55

@ArashVahdat

Arash Vahdat

2 years

📢 Denoising diffusion GAN has been accepted to #ICLR2022 as a spotlight paper 🎉 w/ the amazing Zhisheng Xiao & @karsten_kreis 📢 I will present this work on Friday at @ml_collective reading group organized by @savvyRL . Sign up if you're interested:

@ArashVahdat

Arash Vahdat

2 years

🔥Denoising Diffusion GAN🔥 We tackle the 𝗴𝗲𝗻𝗲𝗿𝗮𝘁𝗶𝘃𝗲 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 𝘁𝗿𝗶𝗹𝗲𝗺𝗺𝗮 with the novel 𝗱𝗲𝗻𝗼𝗶𝘀𝗶𝗻𝗴 𝗱𝗶𝗳𝗳𝘂𝘀𝗶𝗼𝗻 𝗚𝗔𝗡, a diffusion model designed for fast sampling. Zhisheng Xiao @karsten_kreis ft. Peanut 🐈

5

60

299

1

7

55

@ArashVahdat

Arash Vahdat

2 years

Iranian women are cutting their hair in front of the camera to protest the death of #Mahsa_Amini who was killed during a "morality" police arrest in Iran for not having a "proper" hijab. 💔

3

9

54

@ArashVahdat

Arash Vahdat

3 years

Overhyped paper getting citations over & over 😜

@docmilanfar

Peyman Milanfar

3 years

Well known scholar receiving awards

1

8

113

1

2

50

@ArashVahdat

Arash Vahdat

4 years

📢🎉📢 Our work on weakly-supervised phrase grounding has been accepted to #ECCV2020 as a #Spotlight paper. We show how contrastive learning can train an attention model that localizes phrases in an image. Joint work with @tanmay2099 @HoiemDerek @GalChechik @jankautz @visualyang

@tanmay2099

Tanmay Gupta

4 years

Excited to share new work on a contrastive learning formulation for weakly supervised phrase grounding with @ArashVahdat @HoiemDerek @GalChechik @jankautz @visualyang (1/n)

1

6

40

1

6

53

@ArashVahdat

Arash Vahdat

4 years

Setting learning rate is like:

0

2

51

@ArashVahdat

Arash Vahdat

3 years

With an 80GB GPU memory sky's the limit 🤩

@ctnzr

Bryan Catanzaro

3 years

The new 80GB A100 GPU has more than 2 TB/s memory bandwidth! We’ve already started using these @NVIDIAAI to train Megatron language models. 🚀

10

44

324

3

4

52

@ArashVahdat

Arash Vahdat

3 years

Silly question! Is it a bad idea to name a research project such that it's acronym becomes similar to your own name? I'm working on a project that it's acronym naturally becomes VADAT 😁 #GoodProblemsToHave #SillyPhilosophicalQuestions #HowToStartAFriday

10

1

51