# Scaling Vision Transformers Without Training: Prune-and-Merge Token Compression

arXiv: https://arxiv.org/abs/2610.18039 · HTML: https://mlsanity.com/paper/2610.18039/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Token pruning and merging for ViTs, one of 12 directions in Efficient LLM inference and serving.
- Builds on 11 of its 17 references that are in the map.
- First posted on arXiv 13 Oct 2026.

## Placed in

- Efficient LLM inference and serving / [Token pruning and merging for ViTs](https://mlsanity.com/inference/token-pruning-and-merging-for-vits/)

## Builds on (most cited first, 11 of 11)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [Training data-efficient image transformers & distillation through attention](https://mlsanity.com/paper/2012.12877/) (2020)
- [DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification](https://mlsanity.com/paper/2106.02034/) (2021)
- [Not All Patches are What You Need: Expediting Vision Transformers via Token Reorganizations](https://mlsanity.com/paper/2202.07800/) (2022)
- [Which Tokens to Use? Investigating Token Reduction in Vision Transformers](https://mlsanity.com/paper/2308.04657/) (2023)
- [VisionTrim: Unified Vision Token Compression for Training-Free MLLM Acceleration](https://mlsanity.com/paper/2601.22674/) (2026)
- [Adaptive Token Merging for Efficient Transformer Semantic Communication at the Edge](https://mlsanity.com/paper/2509.09955/) (2025)
- [Token Compensator: Altering Inference Cost of Vision Transformer without Re-Tuning](https://mlsanity.com/paper/2408.06798/) (2024)
- [Efficient Token Compression for Vision Transformer with Spatial Information Preserved](https://mlsanity.com/paper/2503.23455/) (2025)
- [Robustness of Probabilistic Models to Low-Quality Data: A Multi-Perspective Analysis](https://mlsanity.com/paper/2512.11912/) (2025)

## Abstract (arXiv)

Vision Transformers achieve strong performance but incur high computational costs from quadratic token interactions. Training-free token reduction methods improve efficiency, but their robustness remains underexplored. We propose Prune-and Merge ViT, a training-free framework for off-the-shelf models. It combines class attention-based pruning with similarity-based merging to reduce redundancy while preserving semantic information. To evaluate robustness we introduce two metrics. Relative Accuracy Drop (RAD) measures relative accuracy degradation under transformations, and Prediction Flip Rate (PFR) quantifies prediction consistency. On pretrained ViT and DeiT, Prune-and-Merge reduces computation competitively while maintaining robustness and better accuracy than existing token reduction methods. Our analysis suggests that hybrid token reduction does not introduce hallucination-like behavior and preserves prediction stability for common image transformations.