Twitxy Saved

Theo Uscidda

@theo_uscidda

Joined October 2023
280 Following 256 Followers 85 Posts

Latest posts 5

Theo Uscidda @theo_uscidda
Check out our work led by @oussamazekri_! Structured noising process (beyond mask and uniform) could be the next step to push the performances of Diffusion LM.
Oussama Zekri Verified account @oussamazekri_ Mar 24
What if discrete diffusion didn鈥檛 have to be stuck with mask or uniform noise? 馃

In our new paper, we show how to go beyond them, unlocking much richer noising processes. And the empirical results are surprisingly strong! 馃殌

馃寪 Project Page: oussamazekri.fr/gdds
馃搼 Paper: arxiv.org/pdf/2603.21342
馃捇 Code: github.com/ozekri/gdds

Thread below 馃У
View quoted post on X
0 replies 2 reposts 9 likes 1.2K views View on X
Theo Uscidda @theo_uscidda
Our work on geometric disentangled representation learning has been accepted to ICLR 2025! 馃帄See you in Singapore if you want to understand this gif better :)
Theo Uscidda @theo_uscidda Dec 14, 2024
Curious about the potential of optimal transport (OT) in representation learning? Join @CuturiMarco's talk at the UniReps workshop today at 2:30 PM! Marco will notably discuss our latest paper on using OT to learn disentangled representations. Details below 猬囷笍
View quoted post on X
0 replies 18 reposts 150 likes 14K views View on X
Theo Uscidda @theo_uscidda
The best resource I鈥檝e found so far on unifying flow matching and diffusion!
Ruiqi Gao Verified account @RuiqiGao Dec 2, 2024
A common question nowadays: Which is better, diffusion or flow matching? 馃

Our answer: They鈥檙e two sides of the same coin. We wrote a blog post to show how diffusion models and Gaussian flow matching are equivalent. That鈥檚 great: It means you can use them interchangeably.
View quoted post on X
0 replies 2 reposts 16 likes 1.6K views View on X