Théophane Vallaeys

@webalorn

PhD student @MetaAI (FAIR Paris) and Sorbonne University | Graduated from ENS | Into generative image modeling

Paris, France webalorn.com Joined May 2023

408 Following 192 Followers 95 Posts

Latest posts 5

Pinned post

Théophane Vallaeys @webalorn Nov 4

Do you really need a video autoencoder for your video generator?

Meet U-JiT: video generation directly in pixel space, trained end-to-end to achieve state-of-the-art results on Kinetics-600

The key: re-thinking compression through the architecture ⬇️

Théophane Vallaeys @webalorn Sep 7

Interested in improving image generation?
Come to our poster at #ECCV2026: "SSDD: Single-Step Diffusion Decoder for Efficient Image Tokenization", on Satudary 12 morning, at ExHall # 65

We show how Diffusion decoders can improve over GAN-based tokenizers for images 🔽

5:00

Théophane Vallaeys @webalorn Oct 7, 2025

🎆 Can we achieve high compression rate for images in autoencoders without compromising quality and decoding speed?
⚡️ We introduce SSDD (Single-Step Diffusion Decoder), achieving improvements on both fonts, setting new state-of-the-art on image reconstruction.

👇 1/N

View quoted post on X

Théophane Vallaeys @webalorn Aug 3

Pixel-space Flow Matching models reach a new level: by delaying the generation of high frequencies, WaiT improves both the coarse structure and the fine-grained details.
Our last work led by @KrunoLehman leads to improvements across image size, model scale and modalities ⬇️

Krunoslav Lehman Pavasovic @KrunoLehman Aug 3

WaiT for the Signal: Simple Frequency-Aware Flow-Matching.

We show how to natively incorporate a fundamental property of images directly into diffusion models; setting a new pixel-space SOTA on ImageNet, while reducing compute.

📄 arxiv.org/abs/2607.28760

Full breakdown below👇

View quoted post on X

Théophane Vallaeys @webalorn Jul 29

Code competitions and studying algorithms teach us how efficiency matters as much as correctness, a subject of importance to me. Training for both with RL is a hard task, as the timing signal is noisy. Yet @PierreChambon6 successfully addresses it here, a must-read: ⬇️

Pierre Chambon Verified account @PierreChambon6 Jul 29

🧵How to combine code correctness and efficiency objectives in online RL?

-> we break the correctness-efficiency Pareto frontier!

+125% relative improvement in CWM32B compared to standard RLVR (13.7 ->30.9 pass@1@top30%)

+150% with Qwen32B

Paper: arxiv.org/abs/2607.25970

View quoted post on X

Théophane Vallaeys @webalorn May 27

Awesome work on pixel diffusion decoders. When making SSDD, I was convinced fast generative non-deterministic decoding was the way to improve latent generation quality, as decoding from lossy latents is a generative task in itself

Xuanchi Ren Verified account @xuanchi13 May 26

The latent-vs-pixel debate misses the point.

GPT Image 2 shows what users notice: pixel-level fidelity.
Latent models show what scales: compact semantic structure.

We connect them by replacing VAE/RAE decoders with a Pixel Diffusion Decoder.

Code and Model available: research.nvidia.com/labs/sil/proje…

🧵(1/N)

View quoted post on X