TensorX
返回文献探索

Paper · arXiv 2606.03746

Qwen-Image-Flash: Beyond Objective Design

Tianhe Wu, Kun Yan, Zikai Zhou, Lihan Jiang, Jiahao Li, Jie Zhang, Kaiyuan Gao, Ningyuan Tang, Shengming Yin, Xiaoyue Chen, Xiao Xu, Yilei Chen, Yuxiang Chen, Yan Shu, Yixian Xu, Yanran Zhang, Zihao Liu, Zhendong Wang, Zekai Zhang, Deqing Li, Liang Peng, Yi Wang, Jingren Zhou, Chenfei Wu

39 upvotesJune 2, 2026arXiv 预印本
AI 摘要

Few-step distillation for visual generative models benefits from systematic investigation of training recipes beyond just distillation objectives, leading to improved student performance through optimized data composition, teacher guidance, and task mixture.

few-step distillationvisual generative modelstext-to-image generationinstruction-guided image editingdistillation objectivestraining recipedata compositionteacher guidancetask mixture

Abstract

Few-step distillation has become an effective strategy for accelerating advanced visual generative models, yet prior work has largely focused on distillation objectives. In this work, we revisit few-step distillation from a complementary perspective, focusing on the training recipe that critically shapes student performance. Using Qwen-Image-2.0 as a representative case, we systematically investigate three factors in unified text-to-image generation and instruction-guided image editing distillation: data composition, teacher guidance, and task mixture. Our empirical analysis reveals several non-obvious behaviors, which motivate the development of Qwen-Image-Flash. Overall, our results suggest that effective few-step distillation requires not only carefully designed objectives, but also principled organization of the broader training pipeline.

北京市昌平区探索星信息技术及软件开发工作室

京ICP备2026059466号
Qwen-Image-Flash: Beyond Objective Design | TensorX