arXiv cs.AI· Canmiao Fu, Chen Li, Guohui Zhang, Hao Lu, Hongyang Wei, Huadong Wu, Huaisong Zhang, Hubery Yin, Liyuan Cui, Qixin Yan, Tong Shao, Xiaoqiang Zhou, Xinhang Leng, Ying Zhang, Zhenhai Liu, Zhipeng Huang·· 12 小时前AI 评分58
WeLM-Image：面向视觉智能的统一图像生成模型
WeLM-Image: A Unified Image Generation Model towards Visual Intelligence
AI 导读
微信团队提出统一图像基础模型 WeLM-Image，把多模态理解融入文生图、指令编辑和多参考图合成，面向微信大规模消费级应用的单 GPU 部署。模型采用 Mixture-of-Transformers 骨干，存储参数 31.7B，每个目标 latent token 约激活 5.4B 生成参数，支持原生 1.5K 分辨率。
正文 · 原文
Authors:Canmiao Fu, Chen Li, Guohui Zhang, Hao Lu, Hongyang Wei, Huadong Wu, Huaisong Zhang, Hubery Yin, Liyuan Cui, Qixin Yan, Tong Shao, Xiaoqiang Zhou, Xinhang Leng, Ying Zhang, Zhenhai Liu, Zhipeng Huang
Abstract:We present \textbf{WeLM-Image}, a unified image foundation model that integrates multimodal understanding into text-to-image generation, instruction-guided editing, and multi-reference synthesis. Designed for single-GPU deployment in WeChat's large-scale consumer applications, the model processes text, reference images, and target latents as one interleaved sequence. A Mixture-of-Transformers (MoT) backbone combines shared attention with independent understanding and generation parameters to limit optimization interference. Equipping each path with Mixture-of-Experts (MoE) modules provides sparse capacity, with 31.7B stored parameters and approximately 5.4B generation parameters activated per target latent token. A generation-oriented data engine combines a knowledge-grounded visual vocabulary with large-scale visual clustering to balance semantic coverage and editing tasks, while training-time evaluation guides iterative data refinement. Bilingual annotations and verified editing supervision support progressive training up to native 1.5K resolution. We optimize generation and editing separately with rubric- and checklist-based rewards, then consolidate the specialized policies through noise-aware on-policy distillation. In the reported overall comparisons, the 1536-resolution model ranks first among the evaluated open-source models on PRISM-Bench, Qwen-Image-Bench, and UnicBench. MeanFlow-based few-step distillation yields WeLM-Image-Turbo, which generates a $1024\times1024$ image in 4.98\,s on a single NVIDIA H20 GPU using eight guidance-free steps.
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2611.07392 [cs.CV] |
| (or arXiv:2611.07392v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2611.07392 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Canmiao Fu [view email]
[v1]
Thu, 5 Nov 2026 03:41:53 UTC (26,708 KB)
来源：arXiv cs.AI · arxiv.org