Twitxy Saved

Amandeep Kumar

@Amandeep__kumar

Ph.D. Student @JohnsHopkins || @AmazonScience Research Fellow
Baltimore, Maryland, US amandpkr.github.io Joined March 2019
839 Following 373 Followers 59 Posts

Latest posts 5

Amandeep Kumar @Amandeep__kumar
Ever wonder why standard Diffusion Transformers fail to converge on high-dimensional representations ? 🤔

Come chat with us about Intrisnic Geometry of the Representations at @eccvconf #ECCV2026 🇸🇪.

📅 Sat, Sep 12 | ⏰ 10:30 AM – 12:30 PM CEST
📍 ExHall # 66

@vishalm_patel
Amandeep Kumar @Amandeep__kumar Feb 11
🚀 Unlocking Standard Diffusion Transformers on Representation Encoders

Why do standard DiTs fail to converge on high-dimensional features like DINOv2? 📉 We found the answer isn't just "more parameters"—it's Geometry.

Introducing Riemannian Flow Matching with Jacobi Regularization (RJF)

📄 Paper: arxiv.org/abs/2602.10099
View quoted post on X
0 replies 13 reposts 107 likes 9.7K views View on X
Amandeep Kumar @Amandeep__kumar
I’m deeply honored and grateful to receive the Amazon AI Ph.D. Fellowship.

Huge thanks to @AmazonScience for their generous support in advancing our research!
Amazon Science Verified account @AmazonScience Oct 21, 2025
🎓 Amazon launches AI PhD Fellowship program, providing $68 million over two years to fund PhD students at 9 universities pursuing research in machine learning, computer vision, and natural-language processing. #AmazonAIFellowship amzn.to/4o6uPlu
View quoted post on X
3 replies 1 reposts 23 likes 4.1K views View on X
Amandeep Kumar @Amandeep__kumar
🧩 “It’s not AR vs diffusion… it’s AR through diffusion.”

👉Is it possible that Visual Autoregressive are models are secretely a discrete diffusion ?

We show: with a Markovian attention mask, VAR becomes mathematically equivalent to discrete diffusion.

Here's how 🧵👇