Ayan Das @dasayan05 Twitter profile

Pinned Tweet

Ayan Das

@dasayan05

1 month

A chapter ends here. 🎓🎓🎓

7

0

60

Last Seen Profiles

@MehendaleR97708

@JeanetteAn54545

@Hockeygirl3784

@SobhanaNair

@PasutriPadang17

@PatriotLiv

@CalamLynch94

@AE86CarGuy

@je6oec

@Hyperobject_Ind

@bryansawyer1379

@PercyBlakeney63

@ViscosoM25

@DAWUK17

@Tantedesijanda

@HareemThim61265

@AkibaRiri3922

@uCA4cQ8vB1SQ5LA

@Anawat80003776

@dislyte

@PasutriPadang17

@RTEToday

@Sofia8202988519

@meganmaguire31

@BngNguy1

@hobitaxi

@deephroact

@Schwede165Wulf

@effysgo

@VollCornHirsch

@20220920_pk3

@SoxsOnn69

@Dellitweets

@FV7hDQvM2t93208

@HumanBeing68502

@NolyangjinJesus

Ayan Das

@dasayan05

5 months

@IntrigueArchive @PicturesFoIder Honestly, this is more American than WWE

4

7

1K

Ayan Das

@dasayan05

3 months

🚨 Diffusion Model's Theory from ground-up 🚨 Accepted at #ICLR2024 BlogPost Track. The blogpost goes back in history and shows the origin of Diffusion Models. It provides a ground-up view of how diffusion model's theory was evolved over time. Link:

Ayan Das

@dasayan05

3 months

Looks like #iclr2024 BlogPost track decisions are out on OpenReview. Aaaand, I got an ACCEPT. 🥰 It was very satisfying writing and submitting content beyond static PDFs. Will share the details soon (Hint: It's on Diffusion Models).

1

0

51

9

122

692

Ayan Das

@dasayan05

2 months

@catshouldnt They come in S, M & L size

0

17

436

Ayan Das

@dasayan05

1 month

@NoContextBrits You must not have seen this one.

2

14

304

Ayan Das

@dasayan05

3 months

@_memeloper what is more funny is it depends on is-odd 🤣

8

0

185

Ayan Das

@dasayan05

8 months

Submitted my PhD Thesis today 🎉. 4 years of work sealed inside a PDF. #phdlife

15

6

135

Ayan Das

@dasayan05

2 years

@fchollet you the same guy right ?

François Chollet

@fchollet

2 years

@ykilcher It was disrespectful to the creator of the repo. "It's just a joke" is not a universal excuse for toxic behavior. Especially not in this context -- not sure if you're aware, but we've been facing a coordinated harassment/insults campaign going on since 2017. Do better.

14

3

127

0

131

Ayan Das

@dasayan05

5 months

@miniapeur Proof by obviousness ?

6

1

96

Ayan Das

@dasayan05

2 months

I can imagine the future. 🤣🤣🤣 (PS: For Indian audience only.)

Gautam Kamath

@thegautamkamath

2 months

NeurIPS 2024 will have a track for papers from high schoolers.

79

91

597

2

8

94

Ayan Das

@dasayan05

6 months

Amazing work by @ZhiyuQu_ , @yizhe_song & others from @SketchXlab on Multi-view Wire Art. They optimise a set of 3D wires such that projection on 3 orthogonal surfaces match given prompts. #generativeart #generativeAI #AIart #sketchart 1/3 🧵

4

22

85

Ayan Das

@dasayan05

2 months

@thegautamkamath Good. I heard NeurIPS Kindergarten Track is coming next year ? 🤣

1

2

82

Ayan Das

@dasayan05

4 months

🚨PhD Defended🚨 Glad to announce that I have successfully defended my PhD. 😁 Examined by Prof. Neill Campbell ( @releasethegecko ) from @UniofBath , and Prof. Richard Bowden from @UniOfSurrey .

9

2

58

Ayan Das

@dasayan05

5 months

@Andr3jH you don't like Ding Liren cuz he has no rizz, Ding Liren doesn't like you cuz you have no ELO

1

54

Ayan Das

@dasayan05

4 months

Started using @huggingface libraries (`diffusers`, `transformers` etc) with PyTorch Lightning ( @LightningAI ). Getting best of both -- top notch support from latest open-source models plus squeaky clean code without boilerplate. Why didn't I do this before ! 🤦 e.g. 👇

1

4

55

Ayan Das

@dasayan05

26 days

Will be at #ICLR2024 in Vienna to present my BlogPost. Tuesday afternoon. Blog Track session 2.

Ayan Das

@dasayan05

3 months

🚨 Diffusion Model's Theory from ground-up 🚨 Accepted at #ICLR2024 BlogPost Track. The blogpost goes back in history and shows the origin of Diffusion Models. It provides a ground-up view of how diffusion model's theory was evolved over time. Link:

9

122

692

0

5

55

Ayan Das

@dasayan05

2 years

#CVPR2022 summary: 1. Boiling temperature at NOLA 2. Reading NeRF posters 3. Searching for @ak92501 4. Reading more NeRF posters & orals 5. … @CVPR

2

51

Ayan Das

@dasayan05

3 months

Looks like #iclr2024 BlogPost track decisions are out on OpenReview. Aaaand, I got an ACCEPT. 🥰 It was very satisfying writing and submitting content beyond static PDFs. Will share the details soon (Hint: It's on Diffusion Models).

1

0

51

Ayan Das

@dasayan05

1 month

@catshouldnt so no one gonna talk about those cat backside fridge magnets

0

46

Ayan Das

@dasayan05

4 months

1/1 paper accepted @iclr_conf . First author Alexander Ashcroft. ❤️ Details to come later. #ICLR2024

2

0

45

Ayan Das

@dasayan05

5 months

I can't explain how much I am enjoying @typstapp . Beautiful, enjoyable, elegant and everything that LaTeX is not. I will surely brag about being an early Typst user to my kids.

1

45

Ayan Das

@dasayan05

8 months

@Rainmaker1973 Thanks for trying to convince me that my ugly pictures are due to focal length. But trust me I am just ugly.

0

1

42

Ayan Das

@dasayan05

2 months

@ShouldHaveCat trying to deceive the world tat he's not orange

0

1

42

Ayan Das

@dasayan05

3 months

My first @LightningAI ⚡️ Studio on using @huggingface 🤗Diffusers with PyTorch Lightning. Make use of SoTA pipelines while having minimal boilerplate. YAML-configurable scripts for training, inference & evaluation.

2

11

38

Ayan Das

@dasayan05

5 months

@sofixspicy @cnviolations no it's not.

0

35

Ayan Das

@dasayan05

10 months

Journey begins. UK -> Hawaii. #ICML2023

1

37

Ayan Das

@dasayan05

24 days

@NoContextBrits Count Binface got 20k some votes. No joke

2

0

33

Ayan Das

@dasayan05

6 months

@kchonyc .. then recruit alex krizhevsky and write AlexNet-v2

0

31

Ayan Das

@dasayan05

9 months

@francoisfleuret @PyTorch How about x[:, k, None] ? Looks more elegant to me.

1

31

Ayan Das

@dasayan05

8 months

@stats_feed Miss universe titles: 🌎 Earth: 71 🌌 Rest of observable universe: 0

1

27

Ayan Das

@dasayan05

6 months

Selected as a top reviewer at @NeurIPSConf ‘23. Glad to contribute. #NeurIPS2023

2

0

28

Ayan Das

@dasayan05

6 months

🐦DreamCreature: Crafting Photorealistic Virtual Creatures from Imagination Check out this cool work from @kam_woh , @EddyZhuxt , @yizhe_song and others that allow you to mix parts to craft imaginary concepts. Paper, code and Playground below. 1/3 🧵

2

7

26

Ayan Das

@dasayan05

3 months

Does @cohere 's ⌘+R work on Windows/Linux ? My keyboard has CTRL key not ⌘. Help, Please.

1

26

Ayan Das

@dasayan05

10 months

@icmlconf poster session in full swing #ICML2023

0

2

25

Ayan Das

@dasayan05

1 year

@karpathy ok, so top researchers use 42 as seed. that's why I was wondering why my model's accuracy is so low. thanks for the pointer, @karpathy

3

0

23

Ayan Das

@dasayan05

2 years

@Google constant 15%? must be decreasing over time, right? if so, how long will it take to achieve ~0% new "never searched before queries"?

7

0

21

Ayan Das

@dasayan05

2 months

@Yoda4ever Bro just wanted show off his fast walking skill

1

0

22

Ayan Das

@dasayan05

1 year

"putting `\vspace{-xx}` just before deadline is an art" ~ Sun Tzu #ICCV2023 @ICCVConference #deadline @CSProfKGD

1

2

21

Ayan Das

@dasayan05

1 year

Paper accepted at @icmlconf 2023. #ICML2023 Details to come later.

4

0

21

Ayan Das

@dasayan05

1 month

@NoContextBrits Gregg Britain

1

21

Ayan Das

@dasayan05

7 months

@angryfermion EigenMarry him

0

19

Ayan Das

@dasayan05

2 years

@savvyRL @ak92501 instead of tweets, he will be handing over latest arXiv paper printouts

0

20

Ayan Das

@dasayan05

5 months

@catshouldnt no way, I almost thought it's a dog

0

1

20

Ayan Das

@dasayan05

10 months

@tomgoldsteincs when your experiments still running but conference deadline is tomorrow

0

20

Ayan Das

@dasayan05

8 months

@karpathy GPT-OS in progress ?

1

18

Ayan Das

@dasayan05

1 year

Glad to announce that I have joined @MediaTek Reseach UK as Senior #DeepLearning Researcher.

7

0

18

Ayan Das

@dasayan05

3 months

@ClementDelangue

1

0

18

Ayan Das

@dasayan05

1 year

Delighted to receive #ICLR2023 acceptance notification while holidaying in a Riad at #Morocco . Can't wait to join @iclr_conf later this year.

1

0

18

Ayan Das

@dasayan05

3 months

@ShouldHaveCat Tiger isn’t Tigerring

0

18

Ayan Das

@dasayan05

2 months

@cnviolations I can now rate it with "high quality source" since it includes the wiki for Dog 🤣

1

0

17

Ayan Das

@dasayan05

1 year

@thegautamkamath Reject. Does not cite and compare with Schmidhuber et al. 1992

0

17

Ayan Das

@dasayan05

23 days

Haha 😛 #ICLR2024 advertisement at Vienna Airport.

1

0

17

Ayan Das

@dasayan05

1 year

The size of the crowd shows the excitement over open AI. At london meetup @weights_biases @StabilityAI

0

1

15

Ayan Das

@dasayan05

8 months

What are the odds that, on a fine evening, the @huggingface hero @RisingSayak would show up in your town ! Thanks for taking some time out of your UK trip. Talked about a lot of things, including Diffusers🧨of course.

0

16

Ayan Das

@dasayan05

4 months

A little after-PhD vacation. Going home after 2 years. 🇬🇧✈️🇮🇳

0

16

Ayan Das

@dasayan05

11 months

This paper by Hyvärinen (2005) is relatively underappreciated. This is where, for the first time, we learned that scores can estimated only from the samples of data distribution. This remarkable result underlies all of modern Diffusion Models.

1

0

15

Ayan Das

@dasayan05

2 years

"kid's crayon drawing on an alien in UFO" #dalle2 by @OpenAI

0

3

14

Ayan Das

@dasayan05

8 months

@darkyudkowsky R for Rizz.

0

15

Ayan Das

@dasayan05

2 months

Phew ! Finished my @icmlconf reviews. Firstly, I got more papers than I could handle and moreover some of the papers are quite content-heavy. I wish the review time is increased in future.

0

1

15

Ayan Das

@dasayan05

7 months

If you are working on GANs and want to name your paper "GANs are all you need", considered abbriviating it as OnlyGANs.

1

0

15

Ayan Das

@dasayan05

8 months

@TheBaldAssassin @NoContextBrits Ya, looks like it was accidentally spilled on the rest

1

0

14

Ayan Das

@dasayan05

2 years

@fchollet REMINDER: Stable Diffusion was originally written in PL and someone had to convert to Keras 🤣

0

15

Ayan Das

@dasayan05

4 months

I keep going back to some papers so much that I even remember their @arxiv ID (i.e. XXXX.YYYYY) Is that normal or should I consult therapy ? 🥲

3

0

14

Ayan Das

@dasayan05

7 months

Our paper, accepted at @NeurIPSConf '23 Diffusion Model Workshop, shows a normalisation trick for DEIS, a SoTA diffusion sampler. Work done primarily by Guoxuan ( @guoxxoug ) during his internship at @MediaTek Research UK. 🧵1/3

2

1

13

Ayan Das

@dasayan05

6 months

🚨DemoFusion: Cool new work by @RuoyiDu , @DL_Chang , @yizhe_song and others that leverage existing Diffusion Models to produce images of increasingly high resolution, without training. Work done in colab with PRIS Lab BUPT, @Tsinghua_Uni , @EdinburghUni and @SketchXlab . 1/4 🧵

1

6

13

Ayan Das

@dasayan05

6 months

I will be at #NeurIPS2023 from 14-16th Dec to present our paper at the Diffusion Workshop on behalf of @guoxxoug and other authors. check the thread for details 👇

Ayan Das

@dasayan05

7 months

Our paper, accepted at @NeurIPSConf '23 Diffusion Model Workshop, shows a normalisation trick for DEIS, a SoTA diffusion sampler. Work done primarily by Guoxuan ( @guoxxoug ) during his internship at @MediaTek Research UK. 🧵1/3

2

1

13

1

0

14

Ayan Das

@dasayan05

3 months

Diffusion paper authors nowadays include every single equations with slightly different notation just to make it look sophisticated. 😓

1

14

Ayan Das

@dasayan05

2 years

@TmlrPub First accepted paper of TMLR. 😍

0

14

Ayan Das

@dasayan05

6 months

Here I come. Me and my poster for #NeurIPS2023

0

14

Ayan Das

@dasayan05

4 months

We are AI folks. We have Tensorboard on 6006 and Jupyter on 8888

Mahsima Dastan

@mahsimadastan

4 months

Are you a localhost:8080 person or localhost:8000 person?

258

14

250

0

2

13

Ayan Das

@dasayan05

5 months

Me: * haven't completed PhD * Journal Editor: "Dear Dr. Ayan ..."

0

13

Ayan Das

@dasayan05

1 year

even if AK is an AI bot, it's a sentient one.

AK

@_akhaliq

1 year

3 am is also a great time to contemplate my life decisions 🤔

2

3

40

1

13

Ayan Das

@dasayan05

10 months

Too much of Generative Modelling in one (few) picture at @icmlconf 2023 diffusion meet-up. #ICML2023 Thanks @sedielem for organising.

0

13

Ayan Das

@dasayan05

2 years

@_akhaliq That's exactly what a bot would say.

0

13

Ayan Das

@dasayan05

10 months

Tomorrow morning I will be flying to #Hawaii for @icmlconf 2023. #ICML2023

Ayan Das

@dasayan05

1 year

Glad to share our @icmlconf 2023 paper, which is now public. In this paper, we derive a “shortest-path” for diffusion model on the distribution space and uncover the corresponding (non-uniform) noising schedule. 👇👇

1

0

12

0

13

Ayan Das

@dasayan05

6 months

Saw this in M3L workshop @NeurIPSConf . #NeurIPS2023

0

13

Ayan Das

@dasayan05

1 year

@francoisfleuret @PyTorch that's easy. first case is simple integer indexing so there is no need for a copy. second case is very general, where a user can request indexes that are far apart -- b[th.tensor([0, 78])] -- due to that it's not possible to return a contiguous "view". so pytorch returns a copy

1

0

12

Ayan Das

@dasayan05

2 months

Finally. Re-designed my personal website -- a little side project. - Built with Quarto (Goodbye Jekyll) - Hosted on Netlify (Bye GitHub Pages) - Focused towards more maintainability and ease of technical publishing (blogs etc) - Commenting, Searching enabled (thanks to Quarto) -

0

12

Ayan Das

@dasayan05

11 months

Phew ! 6 @NeurIPSConf reviews done.

1

0

12

Ayan Das

@dasayan05

1 year

Glad to share our @icmlconf 2023 paper, which is now public. In this paper, we derive a “shortest-path” for diffusion model on the distribution space and uncover the corresponding (non-uniform) noising schedule. 👇👇

Ayan Das

@dasayan05

1 year

Paper accepted at @icmlconf 2023. #ICML2023 Details to come later.

4

0

21

1

0

12

Ayan Das

@dasayan05

2 years

@asrlhhh @_akhaliq because today was ICLR deadline, right

0

1

12

Ayan Das

@dasayan05

4 months

A fine evening at London with friends and supervisor. Amazing Japanese restaurant with good sushi 🍱 and noodles 🍜

0

12

Ayan Das

@dasayan05

2 months

Our favourite image processing test image lena.jpg is now banned by IEEE. 🥲

1

2

12

Ayan Das

@dasayan05

23 days

Arrived.

0

11

Ayan Das

@dasayan05

2 years

@tunguz I heard that tweets containing the word "XGBoost" is being supressed

1

0

10

Ayan Das

@dasayan05

1 year

Majestic view of moon doing gradient descent on hill surface. #Edinburgh #Scotland

0

1

10

Ayan Das

@dasayan05

3 months

The blogpost track of #ICLR2024 is commendable as it allows authors to express a lot more than with static PDFs. This blogpost contains dynamic visualisations that may help the readers. PS: Codes needed for reproducing the visualisation and experiments are provided with the

0

1

10

Ayan Das

@dasayan05

9 months

Recorded version of my talk at @AnalyticsVidhya DataHour is now available at: Slides:

Ayan Das

@dasayan05

9 months

Glad to be invited at the DataHour talk series by @AnalyticsVidhya . On Sep 7, 8:30PM (IST), I'll be speaking at a webinar (free to attend) on "Diffusion Model fundamentals and various applications". Recordings will be available later.

1

0

10

0

1

10

Ayan Das

@dasayan05

6 months

Glad to be back to NOLA for #NeurIPS2023

0

10

Ayan Das

@dasayan05

2 years

@brendanwerth @fchollet that you hate pytorch too 🤣

0

9

Ayan Das

@dasayan05

28 days

@roydanroy success rate 1/PI ?

1

0

10

Ayan Das

@dasayan05

1 year

Community, please let LaTeX retire now. It is too old. Let support new generation tools. @typstapp

EA Aguilar

@EdAlexAguilar

1 year

Looks like a great alternative for LaTeX :) The equations rendered look beautiful, without having to type monsters like \frac{1+\sqrt{5}}{2}. Furthermore, it seems that simple calculations can be done from within + use of variables for effortlessly trying out different outputs.

0

5

36

0

10

Ayan Das

@dasayan05

6 months

@miniapeur I make machines do stuff that humans can do easily

0

10

Ayan Das

@dasayan05

2 years

@bodonoghue85 schmidhuber wants to know your location

0

10

Ayan Das

@dasayan05

5 months

@catshouldnt you mean air-pawed ?

1

0

10

Ayan Das

@dasayan05

7 months

@yong_zhengxin @xwang_lk No, that is exactly called "not solved". This question has a fixed right answer and if it succeeds only sometimes, it is basically failing to reason. Of course figuring out correct seed would make it reproducible, but that's not the point here.

2

0

10

Ayan Das

@dasayan05

9 months

Glad to be invited at the DataHour talk series by @AnalyticsVidhya . On Sep 7, 8:30PM (IST), I'll be speaking at a webinar (free to attend) on "Diffusion Model fundamentals and various applications". Recordings will be available later.

DataHour: Diffusion Model Fundamentals and Various Applications | Analytics Vidhya

[Diffusion Model Fundamentals and Various Applications]🌟 Get Ready for a Mind-Blowing DataHour Extravaganza: Unleash the Magic of Diffusion Models! 🌟 Date: September 7th, 2023 Time: 8:30 PM - 9:30...

community.analyticsvidhya.com

1

0

10

Ayan Das

@dasayan05

2 years

@e2v_sde_parody @CVPR @CSProfKGD @CVPRConf Siraj Raval blesses you, son

0

8

Ayan Das

@dasayan05

2 months

Very interesting. Mixture models with negative mixing weights.

antonio vergari - hiring postdocs & PhD students

@tetraduzione

2 months

Classical mixture models are limited to positive weights and this requires learning very large mixtures! Can we learn (deep) mixtures with negative weights? Answer in our #ICLR2024 spotlight by @loreloc_ Aleks, Martin, Stefan, Nicolas @arnosolin 📜

6

46

216

0

10

Ayan Das

@dasayan05

2 years

@CVPR Will the reviews from withdrawn papers be considered for rating reviewers? If not, it’s sad. I spent so much time on some reviews which were withdrawn. @CVPR

1

0

9