Research map / Architectures and efficiency / Attention and transformers
X-Pruner: eXplainable Pruning for Vision Transformers
Lu Yu, Wei Xiang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use class-wise explainability-aware masks instead of a class-agnostic mask (ablation) | X-Pruner on Swin-T · ILSVRC-12 | Top-1 accuracy drop (%): 2.65 -> 0.51 | Improved |
| add the smoothness regularizer (ablation) | X-Pruner on Swin-T · ILSVRC-12 | Top-1 accuracy drop (%): 1.02 -> 0.51 | Improved |
| add the sparsity regularizer (ablation) | X-Pruner on Swin-T · ILSVRC-12 | Top-1 accuracy drop (%): 1.92 -> 0.51 | Improved |
| learn adaptive layer-wise pruning rates instead of using uniform rates | DeiT-S · ILSVRC-12 | Top-1 accuracy drop (%): 4.05 -> 0.87 | Improved |
| learn adaptive layer-wise pruning rates instead of random pruning | DeiT-S · ILSVRC-12 | Top-1 accuracy drop (%): 2.28 -> 0.87 | Improved |
| apply X-Pruner rather than WDPruning | DeiT-T · ILSVRC-12 | Top-1 accuracy (%): 70.3 -> 71.1 | Improved |
| apply X-Pruner rather than WDPruning | Swin-S · ILSVRC-12 | Top-1 accuracy: 81.8 -> 82.0 | Improved |
| apply X-Pruner rather than ViT-Slim | Swin-T · ILSVRC-12 | FLOPs: 3.4 -> 3.2 | Improved |
Later papers that compared with this method
7 later accepted papers report X-Pruner as a baseline in their results tables; 6 of them report a better number on most of the results they share with it.
- Vulcan: Crafting Compact Class-Specific Vision Transformers For Edge Intelligence (ICLR 2026 · better on 5 of 5 shared results)
- NuWa: Deriving Lightweight Class-Specific Vision Transformers for Edge Devices (CVPR 2026 · better on 3 of 5 shared results)
- Stratified Knowledge-Density Super-Network for Scalable Vision Transformers (AAAI 2026 · better on 1 of 1 shared result)
- Adaptive Pruning of Pretrained Transformer via Differential Inclusions (ICLR 2025 · better on 1 of 1 shared result)
- UPDP: A Unified Progressive Depth Pruner for CNN and Vision Transformer (AAAI 2024 · better on 1 of 1 shared result)
- Adaptive Depth Networks with Skippable Sub-Paths (NeurIPS 2024 · better on 0 of 1 shared result)
- Isomorphic Pruning for Vision Models (ECCV 2024 · better on 1 of 1 shared result)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Zero-TPrune: Zero-Shot Token Pruning through Leveraging of the Attention Graph in Pre-Trained Transformers (CVPR 2024 · similarity 0.91)
- Collaborative Multi-Mode Pruning for Vision-Language Models (CVPR 2026 · similarity 0.88)
- CORP: Closed-Form One-shot Representation-Preserving Structured Pruning for Transformers (NeurIPS 2026 · similarity 0.87)
- UPDP: A Unified Progressive Depth Pruner for CNN and Vision Transformer (AAAI 2024 · similarity 0.87)
- Resource-Efficient Transformer Pruning for Finetuning of Large Models (CVPR 2024 · similarity 0.87)
- Adaptive Pruning of Pretrained Transformer via Differential Inclusions (ICLR 2025 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
