{"arxiv_id":"2611.16345","title":"DRAMA: Dissecting Attention Redundancy for Accelerating Multimodal Diffusion Large Language Models","abstract":"Multimodal diffusion large language models (dLLMs) generate text through iterative masked denoising with bidirectional attention, offering a compelling alternative to autoregressive vision-language modeling. However, each denoising step requires a full forward pass over long multimodal sequences in which visual tokens dominate the sequence length, and a fixed step budget further compounds this quadratic per-step cost. Existing inference acceleration approaches for multimodal dLLMs characterize attention redundancy implicitly, motivating designs from isolated attention visualizations rather than from measurement. We close this gap with DRAMA, a training-free acceleration framework grounded in a systematic dissection of attention redundancy in multimodal dLLMs. By probing attention tensors along spatial and temporal axes, we reveal two persistent redundancy structures: (i) attention mass consistently concentrates on a sparse subset of visual tokens, and (ii) attention distributions rapidly stabilize across denoising steps after an initial formation phase. These findings motivate two complementary mechanisms: Spatial-Redundancy-Aware Visual Token Pruning (SVTP) and Sufficiency-Guided Adaptive Stopping (SGAS). Together, DRAMA achieves an average 18.67 times inference speedup across six vision-language benchmarks without training, architectural modification, or attention caching, while maintaining competitive accuracy. Our results establish a strong, measurement-grounded acceleration baseline for multimodal dLLM inference.","primary_cat":"cs.LG","cross_cats":["cs.AI"],"submitted_at":"2026-11-12T06:04:08+00:00","latest_version":1,"source":"arxiv","source_label":"arXiv","source_url":null,"has_pdf":true,"has_source":false,"verdict":"CONDITIONAL","confidence":"MODERATE","one_line_summary":"A training-free framework prunes visual tokens to one salient patch and stops denoising early, achieving 18.67x faster multimodal dLLM inference at -3.4 pp average accuracy.","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","upvotes":0,"downvotes":0,"pith_score":0,"inbound_count":0,"_links":{"html":"https://pith.science/paper/2611.16345","verdict":"/api/v1/verdicts/2611.16345"}}