# HiAP: A Multi-Granular Stochastic Auto-Pruning Framework for Vision Transformers

arXiv: https://arxiv.org/abs/2603.12222 · HTML: https://mlsanity.com/paper/2603.12222/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 31 of its 38 references that are in the map.
- First posted on arXiv 12 Mar 2026.

## Builds on (most cited first, 30 of 31)

- [An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale](https://mlsanity.com/paper/2010.11929/) (2020)
- [FlashAttention: Fast and Memory-Efficient Exact Attention with IO-Awareness](https://mlsanity.com/paper/2205.14135/) (2022)
- [Deep Compression: Compressing Deep Neural Networks with Pruning, Trained Quantization and Huffman Coding](https://mlsanity.com/paper/1510.00149/) (2015)
- [Token Merging: Your ViT But Faster](https://mlsanity.com/paper/2210.09461/) (2022)
- [DynamicViT: Efficient Vision Transformers with Dynamic Token Sparsification](https://mlsanity.com/paper/2106.02034/) (2021)
- [Are Sixteen Heads Really Better than One?](https://mlsanity.com/paper/1905.10650/) (2019)
- [ProxylessNAS: Direct Neural Architecture Search on Target Task and Hardware](https://mlsanity.com/paper/1812.00332/) (2018)
- [Learning Sparse Neural Networks through L0 Regularization](https://mlsanity.com/paper/1712.01312/) (2017)
- [Importance Estimation for Neural Network Pruning](https://mlsanity.com/paper/1906.10771/) (2019)
- [Reducing Transformer Depth on Demand with Structured Dropout](https://mlsanity.com/paper/1909.11556/) (2019)
- [Not All Patches are What You Need: Expediting Vision Transformers via Token Reorganizations](https://mlsanity.com/paper/2202.07800/) (2022)
- [AdaViT: Adaptive Tokens for Efficient Vision Transformer](https://mlsanity.com/paper/2112.07658/) (2021)
- [DepGraph: Towards Any Structural Pruning](https://mlsanity.com/paper/2301.12900/) (2023)
- [Chasing Sparsity in Vision Transformers: An End-to-End Exploration](https://mlsanity.com/paper/2106.04533/) (2021)
- [Conditional Computation in Neural Networks for faster models](https://mlsanity.com/paper/1511.06297/) (2015)
- [SPViT: Enabling Faster Vision Transformers via Soft Token Pruning](https://mlsanity.com/paper/2112.13890/) (2021)
- [SCOP: Scientific Control for Reliable Neural Network Pruning](https://mlsanity.com/paper/2010.10732/) (2020)
- [Vision Transformer Pruning](https://mlsanity.com/paper/2104.08500/) (2021)
- [Unified Visual Transformer Compression](https://mlsanity.com/paper/2203.08243/) (2022)
- [Global Vision Transformer Pruning with Hessian-Aware Saliency](https://mlsanity.com/paper/2110.04869/) (2021)
- [X-Pruner: eXplainable Pruning for Vision Transformers](https://mlsanity.com/paper/2303.04935/) (2023)
- [Vision Transformer Slimming: Multi-Dimension Searching in Continuous Optimization Space](https://mlsanity.com/paper/2201.00814/) (2022)
- [DBP: Discrimination Based Block-Level Pruning for Deep Model Acceleration](https://mlsanity.com/paper/1912.10178/) (2019)
- [Peeling the Onion: Hierarchical Reduction of Data Redundancy for Efficient Vision Transformer Training](https://mlsanity.com/paper/2211.10801/) (2022)
- [Isomorphic Pruning for Vision Models](https://mlsanity.com/paper/2407.04616/) (2024)
- [Layer Pruning via Fusible Residual Convolutional Block for Deep Neural Networks](https://mlsanity.com/paper/2011.14356/) (2020)
- [LPViT: Low-Power Semi-structured Pruning for Vision Transformers](https://mlsanity.com/paper/2407.02068/) (2024)
- [SR-init: An interpretable layer pruning method](https://mlsanity.com/paper/2303.07677/) (2023)
- [GOHSP: A Unified Framework of Graph and Optimization-based Heterogeneous Structured Pruning for Vision Transformer](https://mlsanity.com/paper/2301.05345/) (2023)
- [UPDP: A Unified Progressive Depth Pruner for CNN and Vision Transformer](https://mlsanity.com/paper/2401.06426/) (2024)

## Abstract (arXiv)

Vision Transformers require significant computational resources and memory bandwidth, severely limiting their deployment on resource-constraint hardware. Most structured pruning methods reduce theoretical cost effectively, yet they typically operate at a single structural granularity and depend on multi-stage pipelines with importance ranking, auxiliary solvers or post-hoc magnitude thresholding, followed by a separate fine-tuning phase to recover accuracy. We propose Hierarchical Auto-Pruning (HiAP), which casts ViT pruning as a single budget-aware learning problem and jointly allocates sparsity across four granularities in one end-to-end phase. HiAP introduces stochastic Gumbel-Sigmoid gates at macro level (attention heads and FFN blocks) and micro level (intra-head dimensions and FFN neurons), and trains them against the task loss together with an analytical MAC cost term. The budget coefficient steers the network to a target compute level while the gates gradually harden into a dense, smaller sub-network at convergence. It does not require importance heuristics, ranking metrics, auxiliary solvers or secondary fine-tuning. On ImageNet, HiAP compresses DeiT-Base to 7.4G MACs at 80.88% top-1 and DeiT-Small to 3.1G at 79.33%, competitive with substantially more complex pipelines at matched compute. The structurally pruned network can be accelerated natively on stock kernels, and more than 90% of the theoretical MAC reduction is realized as measured throughput on an A100.