RoRA: Role-Oriented Regional Allocation for Visual Token Pruning in MLLMs

• URL: http://arxiv.org/abs/2608.07088v1
• Date: Fri, 07 Aug 2026 10:39:47 GMT
• Title: RoRA: Role-Oriented Regional Allocation for Visual Token Pruning in MLLMs
• Abstract summary: We present RoRA, a training-free framework that casts visual token pruning as role-oriented regional evidence allocation.<n>Under matched budgets, RoRA consistently outperforms strong training-free baselines across LLaVA and Qwen-VL families.<n>At a 66.7% pruning ratio, RoRA requires only 0.7 ms for token selection and reduces end-to-end inference time by 24.6%.
• Score: 50.93650157115467
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Multimodal large language models (MLLMs) encode images as long visual token sequences, making prefilling and KV-cache storage expensive. Existing training-free pruning methods select tokens by importance, diversity, or spatial coverage, but treat retained tokens as interchangeable and do not explicitly track which object-related regions are already covered. We present RoRA, a training-free framework that casts visual token pruning as role-oriented regional evidence allocation. Given a fixed budget, RoRA partitions tokens into a protected semantic core, complementary context, and fine-grained detail. It first calibrates text-conditioned attention with a positional prior and a prompt-calibrated object prior, then builds Attention-Anchored Regions (AARs) from high-confidence anchors as lightweight proxies for covered object support. Context is explored mainly outside AARs, while a small AAR-guided budget restores local detail; pairwise similarity is used only for context-stage redundancy filtering. Under matched budgets, RoRA consistently outperforms strong training-free baselines across LLaVA and Qwen-VL families, retaining most of the unpruned accuracy even at aggressive pruning ratios, e.g., 96.5% of full performance at 88.9% pruning on LLaVA-1.5, and improving over D2Pruner by about 5% on Qwen3-VL at 75-90% pruning. At a 66.7% pruning ratio, RoRA requires only 0.7 ms for token selection and reduces end-to-end inference time by 24.6%, corresponding to a 1.33x speedup over unpruned inference on an NVIDIA H800.

Related papers

• ET-Prune: Evidence-Aware Dynamic Budgeting for Visual Token Pruning in Text-Rich MLLMs [37.20525993006875]
  We present ET-Prune, a training-free framework that casts pruning as evidence allocation.<n>It derives question-conditioned evidence from a decoder-side partial query-key block.<n>It safeguards text-like spatial regions, and converts evidence uncertainty and density into a sample-specific token floor.
  arXiv  Detail & Related papers  (2026-08-03T09:42:34Z)
• Beyond Accuracy: Auditing Spatial Provenance in Visual Token Pruning for OCR-Critical MLLM Inference [22.335515110971475]
  Visual-token pruning is judged by answer quality at a fixed retention budget.<n>We turn this blind spot into an evidence-risk audit that couples answer behavior with geometric token-origin provenance.<n>We show that favorable target-verification points do not imply task-general compression.
  arXiv  Detail & Related papers  (2026-07-29T13:45:31Z)
• PP-OCRv6: From 1.5M to 34.5M Parameters, Surpassing Billion-Scale VLMs on OCR Tasks [15.973951643113963]
  PP-OCRv6 is a lightweight OCR system that combines architectural innovation with data-centric optimization.<n>On our in-house benchmarks, PP-OCRv6_medium achieves 83.2% recognition accuracy and 86.2% detection Hmean.
  arXiv  Detail & Related papers  (2026-06-11T09:35:16Z)
• Evading Visual Aphasia: Contrastive Adaptive Semantic Token Pruning for Vision-Language Models [52.78477729846771]
  We introduce COAST (COntrastive Adaptive Semantic Token Pruning), a training-free pruning framework that casts compression as adaptive semantic routing.<n> COAST consistently outperforms strong pruning baselines across token budgets and generalizes across multiple LVLM families.
  arXiv  Detail & Related papers  (2026-05-10T09:07:04Z)
• Prune Redundancy, Preserve Essence: Vision Token Compression in VLMs via Synergistic Importance-Diversity [54.95089105944234]
  Vision-language models (VLMs) face significant computational inefficiencies caused by excessive generation of visual tokens.<n>We propose PruneSID, a training-free Synergistic Importance-Diversity approach featuring a two-stage pipeline.<n>We demonstrate state-of-the-art performance, achieving 96.3% accuracy on LLaVA-1.5 with only 11.1% token retention, and 92.8% accuracy at extreme compression rates (5.6%) on LLaVA-NeXT.
  arXiv  Detail & Related papers  (2026-03-10T10:31:58Z)
• IDPruner: Harmonizing Importance and Diversity in Visual Token Pruning for MLLMs [11.254129271889035]
  Visual token pruning has emerged as a critical technique for accelerating MLLM inference.<n>IDPruner achieves state-of-the-art performance and superior generalization across diverse architectures and tasks.
  arXiv  Detail & Related papers  (2026-02-10T11:20:24Z)
• FASA: Frequency-aware Sparse Attention [56.26881872333624]
  We propose FASA, a novel framework that achieves query-aware token eviction by dynamically predicting token importance.<n>Our key finding is that a small, identifiable subset of "dominant" FCs consistently exhibits high contextual agreement with the full attention head.<n>Across a spectrum of long-context tasks, FASA consistently outperforms all token-eviction baselines and achieves near-oracle accuracy.
  arXiv  Detail & Related papers  (2026-02-03T06:09:06Z)
• D2Pruner: Debiased Importance and Structural Diversity for MLLM Token Pruning [49.16227597771663]
  D2Pruner is a framework that combines debiased importance with a structural pruning mechanism.<n>It reduces FLOPs by 74.2% while retaining 99.2% of its original performance.<n>It marks a significant advancement with up to 63. 53% improvement over existing methods.
  arXiv  Detail & Related papers  (2025-12-22T14:42:31Z)
• MapReduce LoRA: Advancing the Pareto Front in Multi-Preference Optimization for Generative Models [52.32146943039743]
  Reinforcement learning from human feedback has advanced alignment of generative models to human aesthetic and perceptual preferences.<n>We introduce two complementary methods: MapReduce LoRA and Reward-aware Token Embedding (RaTE)<n>Our framework sets a new state-of-the-art multi-preference alignment recipe across modalities.
  arXiv  Detail & Related papers  (2025-11-25T18:49:21Z)
• Patch-Level Contrasting without Patch Correspondence for Accurate and Dense Contrastive Representation Learning [79.43940012723539]
  ADCLR is a self-supervised learning framework for learning accurate and dense vision representation. Our approach achieves new state-of-the-art performance for contrastive methods.
  arXiv  Detail & Related papers  (2023-06-23T07:38:09Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.