AI 译本 · Equalang
新上架PDF · 2026年11月翻译
加载中…
加载中…
arXiv · 2026
WeLM-Image：迈向视觉智能的统一图像生成模型
Canmiao Fu, Chen Li, Guohui Zhang, Hao Lu et al.
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.
我们推出了\textbf{WeLM-Image}，这是一款统一的图像基础模型，将多模态理解融入文生图、指令引导的图像编辑和多参考图合成。该模型面向微信大规模消费级应用中的单 GPU 部署而设计，可将文本、参考图像和目标潜变量作为一个交错序列进行处理。混合变换器（MoT）骨干网络结合了共享注意力机制，以及相互独立的理解与生成参数，从而减少优化干扰。为各个路径配备混合专家（MoE）模块，可提供稀疏容量：模型存储参数量为 31.7B，每个目标潜变量 token 激活的生成参数约为 5.4B。面向生成的数据引擎将基于知识的视觉词汇与大规模视觉聚类相结合，以平衡语义覆盖范围和编辑任务，同时通过训练期间的评估指导数据的迭代优化。双语标注和经过验证的编辑监督数据支持逐步训练，最高可达原生 1.5K 分辨率。我们分别使用基于评分标准和检查清单的奖励来优化生成与编辑，然后通过感知噪声的在线策略蒸馏整合专用策略。在报告的综合比较中，1536 分辨率模型在 PRISM-Bench、Qwen-Image-Bench 和 UnicBench 上均位列受评开源模型之首。基于 MeanFlow 的少步蒸馏得到了 WeLM-Image-Turbo；该模型在单张 NVIDIA H20 GPU 上使用八个无引导步骤，可在 4.98\,s 内生成一张 $1024\times1024$ 图像。
PDF · 2026年11月翻译