# Vision Transformers Need Better Token Interaction

arXiv: https://arxiv.org/abs/2605.23868 · HTML: https://mlsanity.com/paper/2605.23868/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 14 of its 19 references that are in the map.
- First posted on arXiv 22 May 2026.

## Builds on (most cited first, 14 of 14)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [Swin Transformer: Hierarchical Vision Transformer using Shifted Windows](https://mlsanity.com/paper/2103.14030/) (2021)
- [ImageNet Large Scale Visual Recognition Challenge](https://mlsanity.com/paper/1409.0575/) (2014)
- [DINOv2: Learning Robust Visual Features without Supervision](https://mlsanity.com/paper/2304.07193/) (2023)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [The Cityscapes Dataset for Semantic Urban Scene Understanding](https://mlsanity.com/paper/1604.01685/) (2016)
- [Semantic Understanding of Scenes through the ADE20K Dataset](https://mlsanity.com/paper/1608.05442/) (2016)
- [Vision Transformers Need Registers](https://mlsanity.com/paper/2309.16588/) (2023)
- [Sparse Sequence-to-Sequence Models](https://mlsanity.com/paper/1905.05702/) (2019)
- [Adaptively Sparse Transformers](https://mlsanity.com/paper/1909.00015/) (2019)
- [SparseViT: Revisiting Activation Sparsity for Efficient High-Resolution Vision Transformer](https://mlsanity.com/paper/2303.17605/) (2023)
- [Systematic Outliers in Large Language Models](https://mlsanity.com/paper/2502.06415/) (2025)
- [AdaSplash: Adaptive Sparse Flash Attention](https://mlsanity.com/paper/2502.12082/) (2025)
- [AdaSplash-2: Faster Differentiable Sparse Attention](https://mlsanity.com/paper/2604.15180/) (2026)

## Abstract (arXiv)

Vision Transformers (ViTs) can learn strong image-level representations while their patch representations become less effective for dense prediction during prolonged training. We revisit this dense degradation phenomenon and argue that it is not fully explained by high-norm artifacts alone. Instead, we characterize semantic diffusion: an optimization shortcut in which global semantic information spreads through patch tokens beyond what is locally justified. Our analysis shows that dense representation quality is not captured by locality alone: shallow features can remain better aligned with foreground regions yet underperform deeper features, and texttt[CLS] features remain complementary for dense prediction. These observations suggest that the goal should not be to remove global context, but to make token interactions more selective. We therefore study sparse attention as a minimal intervention, replacing softmax attention with entmax-1.5 while preserving global token connectivity. On DINOv1 ViT-S/16 trained for 200 epochs on ImageNet-1K, this change preserves ImageNet linear probing accuracy and substantially improves semantic segmentation performance: VOC mIoU increases from 42.80 to 48.78, ADE20K from 19.85 to 21.97, and Cityscapes from 36.79 to 37.87. These results suggest that selective token mixing is a simple and effective bias for improving dense ViT representations.