Edge-RecViT：基于语义精炼动态递归的高效视觉 Transformer
Edge-RecViT: Efficient Vision Transformer via Semantic-Refined Dynamic Recursion
把「边缘词元富含结构信息」这一先验引入词元级动态深度分配，并配合隐层全共享参数的递归 Transformer，可在不改架构族的前提下同时压缩浮点运算量与参数量，适合直接迁移到 ViT 骨干的轻量化改造。
原文依据 structurally informative edge tokens receive deeper refinement, whereas redundant low-information tokens exit early
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论视觉 Transformer（ViT）在视觉与多模态任务中已取得显著进展，但其部署成本依然高昂。词元自适应方法通过动态深度计算降低浮点运算量，但存在两点局限：（1）全局注意力过度强调高度相似的前景区域，导致词元自适应模块将最深层的计算分配给语义较弱的前景词元，而过早退出富含结构线索的边缘词元（如图 1 所示）；（2）尽管词元自适应降低了浮点运算量，它仍依赖庞大的参数集，且由于词元提前退出，深层权重未被充分利用。参数共享可以缓解冗余，但难以应用于 ViT，因为层次化抽象通常需要多样化的变换。为解决这些问题，我们提出 Edge-RecViT，一种边缘自适应动态递归视觉 Transformer，它将边缘感知的词元自适应排序器与隐层参数完全共享的递归 Transformer 相结合。Edge-RecViT 依据语义丰富度动态分配计算：结构信息丰富的边缘词元获得更深的精炼，而冗余的低信息词元提前退出。大量实验表明，Edge-RecViT 在准确率、浮点运算量与参数效率之间提供了出色的权衡。在 ImageNet-1K 上，其 Top-1 准确率与 DeiT 相差不超过 0.3%，同时将浮点运算量降低 30.5%（从 35.1 降至 24.39 GFLOPs）。在 Base 级别，参数量从 86M 降至 23.21M，且准确率高于 ViT-Base；与 ViT-Large 相比，参数量减少 93%，同时保持更高的准确率。
英文原文摘要 1642 字符
Vision Transformers (ViTs) have achieved remarkable progress in visual and multimodal tasks, yet their deployment remains costly. Token-adaptive methods reduce FLOPs through dynamic depth computation, but they face two limitations: (1) Global attention overemphasizes highly similar foreground regions, causing token-adaptive modules to assign the deepest computation to semantically weak foreground tokens while prematurely exiting edge tokens rich in structural cues (as shown in Fig. 1); (2) Although token-adaption lowers FLOPs, it still relies on large parameter sets, and deep-layer weights remain underutilized due to early token exit. Parameter sharing could address redundancy but is difficult to apply in ViTs, where hierarchical abstraction typically requires diverse transformations. To address these issues, we propose Edge-RecViT, an Edge-Adaptive Dynamic Recursive Vision Transformer that integrates an edge-aware token-adaptive ranker with a recursive transformer using fully shared parameters in its hidden layers. Edge-RecViT dynamically allocates computation based on semantic richness: structurally informative edge tokens receive deeper refinement, whereas redundant low-information tokens exit early. Extensive experiments show that Edge-RecViT provides an excellent trade-off among accuracy, FLOPs, and parameter efficiency. On ImageNet-1K, it matches DeiT within 0.3% Top-1 accuracy while reducing FLOPs by 30.5% (35.1 to 24.39 GFLOPs). At the Base level, parameter drops from 86M to 23.21M with higher accuracy than ViT-Base; compared with ViT-Large, parameters are reduced by 93% while maintaining superior accuracy.