LLaDA-Image: Building Strong Image Generators with Fully Open Training Recipes

2026-09-03

AI Summary

本文提出LLaDA-Image框架，通过仅图像预训练和中程训练建立强大的视觉生成先验，并结合优化技术生成高保真度图像，同时在基准测试上取得领先成绩。

Abstract

We introduce LLaDA-Image, a unified framework that pairs a 6B Diffusion Transformer (DiT) trained from scratch with a frozen vision-language understanding module built on the LLaDA2.0-Mini diffusion language model backbone. Instead of relying heavily on paired image-text data from the beginning, we first build a strong visual generative prior through image-only pre-training and mid-training. The generation pipeline comprises 220M samples, 98 of which are real images. For efficient and scalable optimization, we use parameter-free RMSNorm throughout the DiT together with the Muon optimizer. The resulting unified model produces highly photorealistic images while accurately following fine-grained editing instructions. We further distill LLaDA-Image into LLaDA-Image-Turbo, enabling fast inference in 2-4 sampling steps. On Qwen-Image-Bench, LLaDA-Image achieves overall scores of 53.53 and 53.38 on the English and Chinese tracks, respectively, setting a new state-of-the-art among open-source models on both tracks. To support further research on capable and efficient generative models, we release our model weights, training code, and detailed recipes.

Problem

• Image Generation
• Diffusion Transformer
• Vision-Language Understanding
• Fine-Grained Editing

Innovation

• Diffusion Transformer
• vision-language understanding
• image-only pre-training
• RMSNorm
• fast inference