Twitxy Saved

Manuel Madeira

@manuelmlmadeira

PhD student @epfl | trying to boost science via machine learning
Lausanne, Switzerland manuelmlmadeira.github.io Joined May 2012
537 Following 336 Followers 230 Posts

Latest posts 5

Manuel Madeira @manuelmlmadeira
If you're not surprised, you didn't understand it…🙃

Turns out we can train generative models purely through self-conditioning!

Really happy with the outcome of this collaboration! Huge thanks to @nadjushenka_, @DeschenauxAdri, @jdeschena, and the rest of the team! 🙌
Nadezhda Ilieva @nadjushenka_ Oct 22
Can a language model be based purely on self-conditioning? 🤔

Yes! Meet LAW 👋 (a.k.a. Loopholing All the Way!)

🧵1/8
View quoted post on X
1 replies 6 reposts 39 likes 2.7K views View on X
Manuel Madeira @manuelmlmadeira
Say hi to PUMBA! 🐗

Masked diffusion language models are trained one way and sampled another.

Our new paper studies how to make MDM training more aware of the trajectories followed at inference, and how this can lead to more efficient text generation.

Paper link below 🧵
2 replies 24 reposts 66 likes 2.5K views View on X
Manuel Madeira @manuelmlmadeira
Excited to present #DeFoG with @qinym710 at #ICML2025 !

Catch our oral today at 3:30 PM (West Exhibition Hall C) and join us at the poster after (4:30–7:00 PM, East Exhibition Hall A-B #E-3004).

Come chat graphs & generative models!
🚀 Presenting #DeFoG: our discrete flow‑matching framework for graph generation! Catch our #ICML2025 oral presentation today (3:30 – 3:45 PM, in West Exhibition Hall C) and drop by the poster right after (4:30 –7:00).
Come chat graphs & generative models! @manuelmlmadeira
View quoted post on X
1 replies 4 reposts 11 likes 830 views View on X
Manuel Madeira @manuelmlmadeira
Very happy to see DeFoG accepted as a spotlight at ICML!

We’ve also open-sourced the code.
Looking forward to your feedback on our repo! 🌬️😶‍🌫️
Happy to share that DeFoG: Discrete Flow Matching for Graph Generation will be showcased as a Spotlight Poster at #ICML2025 !

-> Explore the paper: arxiv.org/abs/2410.04263
-> Open-source code: github.com/manuelmlmadeir…

Looking forward to your feedback on our repository!
View quoted post on X
1 replies 2 reposts 19 likes 1.5K views View on X