![[2602.13315] IDPruner: Harmonizing Importance and Diversity in Visual Token Pruning for MLLMs](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.13315] IDPruner: Harmonizing Importance and Diversity in Visual Token Pruning for MLLMs
Summary
The paper presents IDPruner, a novel method for visual token pruning in Multimodal Large Language Models (MLLMs), balancing importance and diversity to enhance computational efficiency without compromising performance.
Why It Matters
As MLLMs grow in complexity, efficient token management becomes crucial for practical applications. IDPruner addresses the challenge of reducing visual token counts while maintaining model performance, making it relevant for researchers and developers in AI and machine learning.
Key Takeaways
- IDPruner achieves a Pareto-optimal balance between token importance and diversity.
- The method does not require attention maps, enhancing compatibility with existing frameworks.
- Extensive experiments show IDPruner retains high performance even with significant token pruning.
Computer Science > Computer Vision and Pattern Recognition arXiv:2602.13315 (cs) [Submitted on 10 Feb 2026] Title:IDPruner: Harmonizing Importance and Diversity in Visual Token Pruning for MLLMs Authors:Yifan Tan, Yifu Sun, Shirui Huang, Hong Liu, Guanghua Yu, Jianchen Zhu, Yangdong Deng View a PDF of the paper titled IDPruner: Harmonizing Importance and Diversity in Visual Token Pruning for MLLMs, by Yifan Tan and 6 other authors View PDF HTML (experimental) Abstract:Multimodal Large Language Models (MLLMs) have demonstrated impressive capabilities, yet they encounter significant computational bottlenecks due to the massive volume of visual tokens. Consequently, visual token pruning, which substantially reduces the token count, has emerged as a critical technique for accelerating MLLM inference. Existing approaches focus on token importance, diversity, or an intuitive combination of both, without a principled framework for their optimal integration. To address this issue, we first conduct a systematic analysis to characterize the trade-off between token importance and semantic diversity. Guided by this analysis, we propose the \textbf{I}mportance and \textbf{D}iversity Pruner (\textbf{IDPruner}), which leverages the Maximal Marginal Relevance (MMR) algorithm to achieve a Pareto-optimal balance between these two objectives. Crucially, our method operates without requiring attention maps, ensuring full compatibility with FlashAttention and efficient deployment via one-shot ...
