Lucas Degeorge
@lucasdegeorge.bsky.social
PhD student at École Polytechnique (Vista) and École des Ponts (IMAGINE) Working on generative models
It's a cool paper, both old school and new school. There's an Easter egg: I designed a game for the teaser. Did you get it right? I tried it in my ML course and almost nobody got it.
Lucas Degeorge@lucasdegeorge.bsky.social · 16h ago🚀 New paper: Balancing Frequencies and Pixels in Flow Matching We tackle the low-frequency bias in pixel-space flow matching and train JiT up to 40% faster without any architectural changes. 📄 Read it here: arxiv.org/abs/2609.02748
🚀 New paper: Balancing Frequencies and Pixels in Flow Matching We tackle the low-frequency bias in pixel-space flow matching and train JiT up to 40% faster without any architectural changes. 📄 Read it here: arxiv.org/abs/2609.02748
🚨 arxiv.org/abs/2604.06129 PoM: A Linear-Time Replacement for Attention with the Polynomial Mixer This paper is the result of doing a lab-wide hackathon on an idea I've had for some time. Probably the paper with the highest number of authors I've ever done. It's a CVPR Findings 26. Thread 🧵👇
PoM: A Linear-Time Replacement for Attention with the Polynomial Mixer
This paper introduces the Polynomial Mixer (PoM), a novel token mixing mechanism with linear complexity that serves as a drop-in replacement for self-attention. PoM aggregates input tokens into a comp...
arxiv.org
👏 Folks! If you are curious about the Generative Modeling via Drifting paper, but you find it difficult to understand → I wrote a different interpretation of it. It's called: "An Expectation-Maximization interpretation of Generative Modeling via Drifting" davidpicard.github.io/pdf/An_Expec...
🎉 Our work MIRO is accepted to #ICML2026 @icmlconf.bsky.social We integrate human preferences directly during pretraining with multi-reward conditioning. ⚡MIRO is 19x faster than baselines and 370x cheaper at inference! 🤗 Try out the models: huggingface.co/spaces/nicol... See you in Seoul 🇰🇷 !
MIRO - a Hugging Face Space by nicolas-dufour
Multi-reward conditioned text-to-image diffusion (ICML 2026)
huggingface.co
Nicolas Dufour@nicolasdufour.bsky.social · 4mo agoThrilled to share that MIRO is accepted to ICML 2026 @icmlconf.bsky.social ! 🎉 By training on the reward scores, we can simply condition the model on high rewards at inference time to guarantee top-tier, aligned outputs. We’ve updated our paper with some additional results!
Thrilled to share that MIRO is accepted to ICML 2026 @icmlconf.bsky.social ! 🎉 By training on the reward scores, we can simply condition the model on high rewards at inference time to guarantee top-tier, aligned outputs. We’ve updated our paper with some additional results!
MIRO: MultI-Reward cOnditioned pretraining improves T2I quality and efficiency
The default paradigm of post-training text-to-image generators includes post-hoc selection of generated images, and subsequent training with one reward model to align the generator to the reward, typi...
arxiv.org
Nicolas Dufour@nicolasdufour.bsky.social · 10mo agoWe introduce MIRO: a new paradigm for T2I model alignment integrating reward conditioning into pretraining, eliminating the need for separate fine-tuning/RL stages. This single-stage approach offers unprecedented efficiency and control. - 19x faster convergence ⚡ - 370x less FLOPS than FLUX-dev 📉
Check out our new work: MIRO No more post-training alignment! We integrate human alignment right from the start, during pretraining! Results: ✨ 19x faster convergence ⚡ ✨ 370x less compute 💻 🔗 Explore the project: nicolas-dufour.github.io/miro/
MIRO: Multi-Reward Conditioning for Efficient Text-to-Image Generation
Train once, align many rewards. MIRO achieves 19× faster convergence and 370× less compute than FLUX while reaching GenEval score of 75. Controllable trade-offs at inference time.
nicolas-dufour.github.io
Nicolas Dufour@nicolasdufour.bsky.social · 10mo agoWe introduce MIRO: a new paradigm for T2I model alignment integrating reward conditioning into pretraining, eliminating the need for separate fine-tuning/RL stages. This single-stage approach offers unprecedented efficiency and control. - 19x faster convergence ⚡ - 370x less FLOPS than FLUX-dev 📉
We introduce MIRO: a new paradigm for T2I model alignment integrating reward conditioning into pretraining, eliminating the need for separate fine-tuning/RL stages. This single-stage approach offers unprecedented efficiency and control. - 19x faster convergence ⚡ - 370x less FLOPS than FLUX-dev 📉
Very proud of our recent work, kudos to the team! Read @davidpicard.bsky.social’s excellent post for more details or the paper arxiv.org/pdf/2502.21318
arxiv.org
David Picard@davidpicard.eurosky.social · 11mo ago🚨Updated: "How far can we go with ImageNet for Text-to-Image generation?" TL;DR: train a text2image model from scratch on ImageNet only and beat SDXL. Paper, code, data available! Reproducible science FTW! 🧵👇 📜 arxiv.org/abs/2502.21318 💻 github.com/lucasdegeorg... 💽 huggingface.co/arijitghosh/...
Final note: I'm (we're) tempted to organize a challenge on that topic as a workshop at a CV conf. ImageNet is the only source of images allowed and then you compete to get the bold numbers. Do you think there would be people in for that? Do you think it would make for a nice competition?
🚨Updated: "How far can we go with ImageNet for Text-to-Image generation?" TL;DR: train a text2image model from scratch on ImageNet only and beat SDXL. Paper, code, data available! Reproducible science FTW! 🧵👇 📜 arxiv.org/abs/2502.21318 💻 github.com/lucasdegeorg... 💽 huggingface.co/arijitghosh/...
I had the privilege to be invited to speak about our work "Around the World in 80 Timesteps" at the French Podcast Underscore! If you speak french, i highly recommend it they did a great job with the montage! If you want to learn more nicolas-dufour.github.io/plonk www.youtube.com/watch?v=s5oH...
Il a conçu la première IA d’OSINT (terrifiant… et génial)
YouTube video by Underscore_
youtube.com
If you want to listen to Nicolas (in French) talking about generative models for geolocation, it's right now: m.twitch.tv/micode
Micode - Twitch
🌐 UNDERSCORE_
m.twitch.tv
🚨 News! 🚨 We have released the models from our latest paper "How far can we go with ImageNet for text-to-image generation?" Check out the models on HuggingFace: 🤗 huggingface.co/Lucasdegeorg... 📜 arxiv.org/abs/2502.21318
Lucasdegeorge/CAD-I · Hugging Face
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
huggingface.co
David Picard@davidpicard.eurosky.social · 2y ago🚨 New preprint! How far can we go with ImageNet for Text-to-Image generation? w. @arrijitghosh.bsky.social @lucasdegeorge.bsky.social @nicolasdufour.bsky.social @vickykalogeiton.bsky.social TL;DR: Train a text-to-image model using 1000 less data in 200 GPU hrs! 📜https://arxiv.org/abs/2502.21318 🧵👇
Text-to-image models are trained on billions of data. But, is it necessary? Our "How far can we go with ImageNet for T2I generation?" @lucasdegeorge.bsky.social @arrijitghosh.bsky.social @nicolasdufour.bsky.social @davidpicard.bsky.social shows that no, if we are careful arxiv.org/abs/2502.21318
David Picard@davidpicard.eurosky.social · 2y ago🚨 New preprint! How far can we go with ImageNet for Text-to-Image generation? w. @arrijitghosh.bsky.social @lucasdegeorge.bsky.social @nicolasdufour.bsky.social @vickykalogeiton.bsky.social TL;DR: Train a text-to-image model using 1000 less data in 200 GPU hrs! 📜https://arxiv.org/abs/2502.21318 🧵👇
Wow, neet! Reannotation is key here. Conjecture: As we are get more and more well-aligned text-image data, it will become easier and easier to train models. This will allow us to explore both more streamlined and more exotic training recipes. More signals that exciting times are coming!
David Picard@davidpicard.eurosky.social · 2y ago🚨 New preprint! How far can we go with ImageNet for Text-to-Image generation? w. @arrijitghosh.bsky.social @lucasdegeorge.bsky.social @nicolasdufour.bsky.social @vickykalogeiton.bsky.social TL;DR: Train a text-to-image model using 1000 less data in 200 GPU hrs! 📜https://arxiv.org/abs/2502.21318 🧵👇
These are some ridiculously good results from training tiny T2I models purely on ImageNet! It's almost too good to be true. Do check it out!
Nicolas Dufour@nicolasdufour.bsky.social · 2y agoCheck out our latest work on Text-to-Image generation! We've successfully trained a T2I model using only ImageNet data by leveraging captioning and data augmentation.
🚨 New preprint! How far can we go with ImageNet for Text-to-Image generation? w. @arrijitghosh.bsky.social @lucasdegeorge.bsky.social @nicolasdufour.bsky.social @vickykalogeiton.bsky.social TL;DR: Train a text-to-image model using 1000 less data in 200 GPU hrs! 📜https://arxiv.org/abs/2502.21318 🧵👇
🌍 Guessing where an image was taken is a hard, and often ambiguous problem. Introducing diffusion-based geolocation—we predict global locations by refining random guesses into trajectories across the Earth's surface! 🗺️ Paper, code, and demo: nicolas-dufour.github.io/plonk
