# CAIT: Triple-Win Compression towards High Accuracy, Fast Inference, and Favorable Transferability For ViTs

arXiv: https://arxiv.org/abs/2309.15755 · HTML: https://mlsanity.com/paper/2309.15755/

## Key facts

- Cited by 9 papers in the map (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 83 of its 120 references that are in the map.
- First posted on arXiv 27 Sep 2023.

## 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, 30 of 83)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Microsoft COCO: Common Objects in Context](https://mlsanity.com/paper/1405.0312/) (2014)
- [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)
- [Segment Anything](https://mlsanity.com/paper/2304.02643/) (2023)
- [Masked Autoencoders Are Scalable Vision Learners](https://mlsanity.com/paper/2111.06377/) (2021)
- [Mask R-CNN](https://mlsanity.com/paper/1703.06870/) (2017)
- [Emerging Properties in Self-Supervised Vision Transformers](https://mlsanity.com/paper/2104.14294/) (2021)
- [PaLM: Scaling Language Modeling with Pathways](https://mlsanity.com/paper/2204.02311/) (2022)
- [Training data-efficient image transformers & distillation through attention](https://mlsanity.com/paper/2012.12877/) (2020)
- [Deformable DETR: Deformable Transformers for End-to-End Object Detection](https://mlsanity.com/paper/2010.04159/) (2020)
- [Emergent Abilities of Large Language Models](https://mlsanity.com/paper/2206.07682/) (2022)
- [BEiT: BERT Pre-Training of Image Transformers](https://mlsanity.com/paper/2106.08254/) (2021)
- [Masked-attention Mask Transformer for Universal Image Segmentation](https://mlsanity.com/paper/2112.01527/) (2021)
- [Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions](https://mlsanity.com/paper/2102.12122/) (2021)
- [Context Encoders: Feature Learning by Inpainting](https://mlsanity.com/paper/1604.07379/) (2016)
- [Rethinking Semantic Segmentation from a Sequence-to-Sequence Perspective with Transformers](https://mlsanity.com/paper/2012.15840/) (2020)
- [Unified Perceptual Parsing for Scene Understanding](https://mlsanity.com/paper/1807.10221/) (2018)
- [Per-Pixel Classification is Not All You Need for Semantic Segmentation](https://mlsanity.com/paper/2107.06278/) (2021)
- [Tokens-to-Token ViT: Training Vision Transformers from Scratch on ImageNet](https://mlsanity.com/paper/2101.11986/) (2021)
- [Token Merging: Your ViT But Faster](https://mlsanity.com/paper/2210.09461/) (2022)
- [Scaling Vision Transformers](https://mlsanity.com/paper/2106.04560/) (2021)
- [Pre-Trained Image Processing Transformer](https://mlsanity.com/paper/2012.00364/) (2020)
- [DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification](https://mlsanity.com/paper/2106.02034/) (2021)
- [CrossViT: Cross-Attention Multi-Scale Vision Transformer for Image Classification](https://mlsanity.com/paper/2103.14899/) (2021)
- [PaLI: A Jointly-Scaled Multilingual Language-Image Model](https://mlsanity.com/paper/2209.06794/) (2022)
- [Scaling Vision with Sparse Mixture of Experts](https://mlsanity.com/paper/2106.05974/) (2021)
- [Scaling Vision Transformers to 22 Billion Parameters](https://mlsanity.com/paper/2302.05442/) (2023)

## Built on by (most cited first, 9 of 9)

- [Efficient Multimodal Large Language Models: A Survey](https://mlsanity.com/paper/2405.10739/) (2024)
- [[CLS] Token Tells Everything Needed for Training-free Efficient MLLMs](https://mlsanity.com/paper/2412.05819/) (2024)
- [PYRA: Parallel Yielding Re-Activation for Training-Inference Efficient Task Adaptation](https://mlsanity.com/paper/2403.09192/) (2024)
- [LSNet: See Large, Focus Small](https://mlsanity.com/paper/2503.23135/) (2025)
- [PrefixKV: Adaptive Prefix KV Cache is What Vision Instruction-Following Models Need for Efficient Generation](https://mlsanity.com/paper/2412.03409/) (2024)
- [Towards Efficient Vision-Language Tuning: More Information Density, More Generalizability](https://mlsanity.com/paper/2312.10813/) (2023)
- [Exploring The Neural Burden In Pruned Models: An Insight Inspired By Neuroscience](https://mlsanity.com/paper/2407.16716/) (2024)
- [ToaSt: Token Channel Selection and Structured Pruning for Efficient ViT](https://mlsanity.com/paper/2602.15720/) (2026)
- [Decompose, Mix, Adapt: A Unified Framework for Parameter-Efficient Neural Network Recombination and Compression](https://mlsanity.com/paper/2603.27383/) (2026)

## Abstract (arXiv)

Vision Transformers (ViTs) have emerged as state-of-the-art models for various vision tasks recently. However, their heavy computation costs remain daunting for resource-limited devices. To address this, researchers have dedicated themselves to compressing redundant information in ViTs for acceleration. However, existing approaches generally sparsely drop redundant image tokens by token pruning or brutally remove channels by channel pruning, leading to a sub-optimal balance between model performance and inference speed. Moreover, they struggle when transferring compressed models to downstream vision tasks that require the spatial structure of images, such as semantic segmentation. To tackle these issues, we propose CAIT, a joint underlinecompression method for ViTs that achieves a harmonious blend of high underlineaccuracy, fast underlineinference speed, and favorable underlinetransferability to downstream tasks. Specifically, we introduce an asymmetric token merging (ATME) strategy to effectively integrate neighboring tokens. It can successfully compress redundant token information while preserving the spatial structure of images. On top of it, we further design a consistent dynamic channel pruning (CDCP) strategy to dynamically prune unimportant channels in ViTs. Thanks to CDCP, insignificant channels in multi-head self-attention modules of ViTs can be pruned uniformly, significantly enhancing the model compression. Extensive experiments on multiple benchmark datasets show that our proposed method can achieve state-of-the-art performance across various ViTs.