Research map / Vision-language and multimodal / Multimodal LLMs
MVPruner: Dynamic Token Pruning for Accelerating Multi-view Vision-Language Models in Autonomous Driving
Nan Yang, Zhanwen Liu, Linfeng Zhang, Shangyu Xie, Yang Wang, Wenzhuo Zhou, Xiangmo Zhao
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 MVPruner instead of Prune2Drive at 90% pruning | DriveLMM-o1 · DriveLMM-o1 | Overall Reasoning: 70.31 -> 72.91 | Improved |
| use MVPruner instead of Prune2Drive at 90% pruning | DriveMM · MAPLM | Average: 64.81 -> 68.30 | Improved |
| use MVPruner instead of Prune2Drive at 90% pruning | DriveMM · DriveLM | Average: 57.4 -> 58.2 | Improved |
| use MVPruner instead of Prune2Drive at 90% pruning | DriveMM · STSnu | Average: 43.17 -> 43.00 | No clear effect |
| use MVPruner at 75% pruning | DriveMM · STSnu | Average: 44.21 -> 44.25 | Improved |
| use MVPruner instead of Prune2Drive at 90% pruning | DriveMM · MAPLM | FRM accuracy: 48.33 -> 53.27 | Improved |
| use MVPruner instead of Prune2Drive | DriveMM · DriveLM efficiency | decoding time (ms): 59.9 -> 51.8 | Improved |
| use MVPruner instead of Prune2Drive | DriveMM · DriveLM efficiency | prefilling time (ms): 307.5 -> 395.6 | Hurt |
| add second-stage pruning with diversity measure kept (Stage 1 alone applied to shallow layers) (ablation) | MVPruner · DriveLM and MAPLM (90% pruning) | accuracy | Improved |
| remove future contribution metric from CCTS (ablation) | MVPruner · DriveLM and MAPLM (90% pruning) | accuracy | Hurt |
| use minimum pairwise distance instead of other diversity metrics in DRA (ablation) | MVPruner · DriveLM and MAPLM (90% pruning) | accuracy | No clear effect |
| place stage 1 at deeper layers (ablation) | MVPruner · DriveLM and MAPLM (90% pruning) | accuracy | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- Prune2Drive — Prune2Drive: A Plug-and-Play Framework for Accelerating Vision-Language Models in Autonomous Driving (CVPR 2026) · better on 10 of 12 shared results
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
- Prune2Drive: A Plug-and-Play Framework for Accelerating Vision-Language Models in Autonomous Driving (CVPR 2026 · similarity 0.95)
- Each Complexity Deserves a Pruning Policy (NeurIPS 2025 · similarity 0.92)
- ATP-LLaVA: Adaptive Token Pruning for Large Vision Language Models (CVPR 2025 · similarity 0.90)
- Balanced Token Pruning: Accelerating Vision Language Models Beyond Local Optimization (NeurIPS 2025 · similarity 0.90)
- IF-Prune: Information-Flow Guided Token Pruning for Efficient Vision-Language Models (CVPR 2026 · similarity 0.90)
- QuietPrune: Query-Guided Early Token Pruning for Vision-Language Models (CVPR 2026 · similarity 0.89)
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.
