Ankit Dhall

263 posts

Ankit Dhall banner
Ankit Dhall

Ankit Dhall

@ankitdhall

Sr. Deep Learning Engineer @NVIDIA Visual Gen AI | DL Algorithms Prev. @latticeflowai Seervision @ETH @amzracing @motionaldrive @UniFreiburg @iiit_hyderabad VIT

Zurich Se unió Aralık 2009
1.7K Siguiendo241 Seguidores
Tweet fijado
Ankit Dhall
Ankit Dhall@ankitdhall·
Building generative models using libraries is great for getting started but the high-level APIs abstract away the details (usually where the devil is) I implemented Torchsmith, an open-source package that implements generative models from scratch using basic PyTorch operations🧵
Ankit Dhall tweet media
English
1
0
0
105
Artem Lukoianov
Artem Lukoianov@ottogin1·
Hi, Ankit, agree, hah! We are figuring out the logistics of recording with @CVPR, even if for some reason we won't be able to record on the conference, we will make sure to reenact it and upload the videos. In the meantime, check out our MIT mini course where we cover some parts of it: practical-diffusion.org
English
1
0
1
61
Ankit Dhall
Ankit Dhall@ankitdhall·
I'm excited to start a new journey at @NVIDIA as Senior Deep Learning Engineer working on Visual Generative AI in the Deep Learning Algorithms team!
English
0
0
1
42
Joey Bose
Joey Bose@bose_joey·
Heads up, NeurIPS attendees! 👋 🚨Don't miss the FPI Workshop (Frontiers of Probabilistic Inference) back again on Dec 7th. 🔉We've curated an incredible set of accepted papers showcasing the future of probabilistic modeling. Schedule & papers: fpineurips.framer.website We also have a stellar lineup of invited speakers that should be of broad interest to people in AI4Science #neurips2025 #ProbabilisticML #AI Rumours are that there's also a workshop party🫢. co-organized with a stellar team @tara_aksa @msalbergo @marylougab @theh2o64 @guanhorng_liu @k_neklyudov Grant Rotskoff @EvaSmorodina @AlexanderTong7
English
4
20
80
14.4K
Ankit Dhall retuiteado
Jia-Bin Huang
Jia-Bin Huang@jbhuang0604·
Scrolling the AI news timeline as a researcher feels like a teenager browsing Instagram: "Everyone else has figured everything out!" Reliable home robots imminent, 100× productivity AI agents, insane visual generation ... Exciting, but anxiety-inducing. What am I doing? 😬
English
35
28
525
32.6K
Ankit Dhall
Ankit Dhall@ankitdhall·
@jm_alexia Was going to request for a recording but you have already taken care of it🙏
English
0
0
1
68
Alexia Jolicoeur-Martineau
Alexia Jolicoeur-Martineau@jm_alexia·
I will give a presentation on "Tiny Recursion Models" tomorrow at 1pm in the Mila Agora (6650 Saint-Urbain, Montreal). Its open to everyone, feel free to come by!
English
18
18
278
75.3K
Ankit Dhall
Ankit Dhall@ankitdhall·
(5/5) This thread is a concise version, check out the repo on GitHub for the full details github.com/ankitdhall/tor… P.S.: Definitely recommend this way for anyone who wants to get a deeper understanding and see how to train and sample from such models
English
0
0
0
31
Ankit Dhall
Ankit Dhall@ankitdhall·
(4/5) Flow matching: - ODE formulation: Gaussian flow matching and Linear flow matching - SDE formulation: Langevin flow matching
Ankit Dhall tweet media
English
1
0
0
47
Ankit Dhall
Ankit Dhall@ankitdhall·
Building generative models using libraries is great for getting started but the high-level APIs abstract away the details (usually where the devil is) I implemented Torchsmith, an open-source package that implements generative models from scratch using basic PyTorch operations🧵
Ankit Dhall tweet media
English
1
0
0
105
Ankit Dhall retuiteado
Pedro Domingos
Pedro Domingos@pmddomingos·
Matrix multiplication is now a multi-trillion dollar business.
English
243
934
9.1K
444.4K
Ankit Dhall retuiteado
Jack Morris
Jack Morris@jxmnop·
it’s been an interesting ride watching the conventional nomenclature of machine learning gradually lose all meaning. there used to be TRAIN and TEST and everything was simple. now we train on the universe. and we test on the universe, too. are we gaming our benchmarks? are we extrapolating or interpolating? if a model is trained on the entire internet but generates a single novel sentence, is it just combining phrases from Reddit or writing something truly novel? feels like we lack the words to even describe the systems we’ve built.
English
42
47
957
100K
Sarath Chandar
Sarath Chandar@apsarathchandar·
I am teaching intro to ML this Fall (starting today!), and watching @3blue1brown linear algebra videos several times is a mandatory prerequisite for the students!
Sarath Chandar tweet media
English
9
34
788
48.6K
Ankit Dhall retuiteado
Kevin Zhang
Kevin Zhang@kevinzhang25·
while doing ML research i spend a disproportionately large amount of time looking at this specific figure
Kevin Zhang tweet media
English
27
78
1.7K
103.1K
Ankit Dhall
Ankit Dhall@ankitdhall·
@giffmana @XiaohuaZhai @__kolesnikov__ @_basilM Okay, so the values were obtained via hyper parameter search. Not clear what you mean when referring to "prior". > "This is because the bias term ensures that the training starts close to the prior"
English
1
0
0
90
Ankit Dhall
Ankit Dhall@ankitdhall·
@giffmana @XiaohuaZhai @__kolesnikov__ @_basilM "We initialize t' and b to log 10 and −10 respectively. This makes sure the training starts roughly close to the prior ..." Couldn't find more about how the two values were determined in the text.
English
1
0
0
62
Lucas Beyer (bl16)
Lucas Beyer (bl16)@giffmana·
What makes CLIP work? The contrast with negatives via softmax? The more negatives, the better -> large batch-size? We'll answer "no" to both in our ICCV oral🤓 By introducing SigLIP, a simpler CLIP that also works better and is more scalable, we can study the extremes. Hop in🧶
Lucas Beyer (bl16) tweet media
English
26
290
1.7K
453.6K