研究人员推出了一种新颖的知识蒸馏框架VarKD，旨在压缩计算密集型的自回归（AR）图像生成模型。研究强调，尽管标准蒸馏方法在语言建模中取得了成功，但由于解码范围长和视觉令牌模糊等挑战，它们对视觉AR模型的有效性较低。VarKD通过在具有选择性教师监督和减少的令牌级模糊的学生样本上进行蒸馏来解决这些问题，并在ImageNet上展示了改进的性能。 AI
影响 VarKD提供了一种更有效的方式来部署强大的视觉AR模型，可能降低计算成本并实现更广泛的可访问性。
排序理由 该集群包含一篇详细介绍模型压缩新方法的学术论文。
AI 生成摘要 · Google Gemini · 来自 2 个来源。 我们如何撰写摘要 →
arXiv:2606.06078v1 Announce Type: new Abstract: Autoregressive (AR) image generation models are highly expressive but computationally intensive, motivating effective model compression. Knowledge distillation (KD) is a natural approach for model compression and has been widely stu…
Autoregressive (AR) image generation models are highly expressive but computationally intensive, motivating effective model compression. Knowledge distillation (KD) is a natural approach for model compression and has been widely studied in language modeling, yet its behavior in v…