Seeing the End at Step Zero: Accelerating Diffusion MLLMs via MLP Sparsity-Aware Truncation

• URL: http://arxiv.org/abs/2607.14557v1
• Date: Thu, 16 Jul 2026 04:37:02 GMT
• Title: Seeing the End at Step Zero: Accelerating Diffusion MLLMs via MLP Sparsity-Aware Truncation
• Abstract summary: Seer is a training-free framework that detects a valid semantic boundary at the very first denoising step through a shift in activation sparsity.<n> Experimental results demonstrate that Seer effectively eliminates padding waste, accelerating throughput by up to $sim$31$times$.
• Score: 6.796779304354568
• License: http://creativecommons.org/licenses/by/4.0/
• Abstract: Diffusion Multimodal Large Language Models (DMLLMs) are highly effective for multimodal reasoning, yet their inference efficiency is significantly hindered by fixed-length generation constraints. Since the actual output length is unknown, output sequences are padded to a predefined maximum length, resulting in substantial redundant computation over unnecessary [EOS] tokens. In this work, we discover that DMLLMs implicitly reveal their valid semantic boundary at the very first denoising step through a distinct shift in MLP activation sparsity. Leveraging this observation, we propose Seer, a training-free framework that detects this boundary using a Signal-to-Noise Ratio (SNR)-based criterion and performs one-shot truncation of the redundant suffix for all subsequent computations. To preserve these theoretical gains during batched serving, Seer incorporates a hybrid execution strategy that maximizes throughput while seamlessly accommodating dynamic sequence lengths. Experimental results demonstrate that Seer effectively eliminates padding waste, accelerating throughput by up to $\sim$31$\times$. Across 9 benchmarks, Seer robustly maintains overall performance and even improves accuracy on complex visual tasks by mitigating noise leakage (e.g., DocVQA score increases from 63.52 to 63.66), offering a highly efficient, plug-and-play solution for DMLLM acceleration.

Related papers

• Constrained Decoding for Diffusion Language Models via Efficient Inference over Finite Automata [57.27430779838529]
  We present an exact and tractable algorithm for sampling from the constrained mean-field posterior under any constraint expressible as a finite automaton.<n>The approach guarantees constraint satisfaction by construction, supports both greedy and sampling-based decoding, and is compatible with parallel and block-wise decoding.<n> Empirical evaluations on Dream-7B and LLaDA-8 show substantial accuracy gains across various tasks.
  arXiv  Detail & Related papers  (2026-07-08T05:48:57Z)
• SVD Contextual Sparsity Predictors for Fast LLM Inference [0.0]
  This paper presents a framework for accelerating inference of ReGLU-based feed-forward networks (FFNs) within large language models (LLMs)<n>Experiments demonstrate up to a 1.8x reduction in end-to-end decoding time while maintaining less than 1% degradation in benchmark scores on tasks involving complex math and code generation.
  arXiv  Detail & Related papers  (2026-03-14T20:36:13Z)
• DyLLM: Efficient Diffusion LLM Inference via Saliency-based Token Selection and Partial Attention [2.7422645382944935]
  We present DyLLM, a training-free inference framework that accelerates decoding by selectively computing only salient tokens.<n>Across diverse reasoning and code-generation benchmarks, DyLLM achieves up to 9.6x higher throughput.
  arXiv  Detail & Related papers  (2026-03-09T07:02:01Z)
• Prism: Efficient Test-Time Scaling via Hierarchical Search and Self-Verification for Discrete Diffusion Language Models [96.0074341403456]
  Inference-time compute has re-emerged as a practical way to improve LLM reasoning.<n>Most test-time scaling (TTS) algorithms rely on autoregressive decoding.<n>We propose Prism, an efficient TTS framework for dLLMs.
  arXiv  Detail & Related papers  (2026-02-02T09:14:51Z)
• Beyond Next-Token Prediction: A Performance Characterization of Diffusion versus Autoregressive Language Models [82.87985794856803]
  Large Language Models (LLMs) have achieved state-of-the-art performance on a broad range of Natural Language Processing (NLP) tasks.<n>Recently, Diffusion Language Models (DLMs) have emerged as a promising alternative architecture.
  arXiv  Detail & Related papers  (2025-10-05T10:50:52Z)
• Beyond Fixed: Training-Free Variable-Length Denoising for Diffusion Large Language Models [74.15250326312179]
  Diffusion Large Language Models offer efficient parallel generation and capable global modeling.<n>The dominant application ofDLLMs is hindered by the need for a statically predefined generation length.<n>We introduce DAEDAL, a novel training-free denoising strategy that enables Dynamic Adaptive Length Expansion.
  arXiv  Detail & Related papers  (2025-08-01T17:56:07Z)
• END: Early Noise Dropping for Efficient and Effective Context Denoising [60.24648712022382]
  Large Language Models (LLMs) have demonstrated remarkable performance across a wide range of natural language processing tasks.<n>They are often distracted by irrelevant or noisy context in input sequences that degrades output quality.<n>We introduce Early Noise Dropping (textscEND), a novel approach to mitigate this issue without requiring fine-tuning the LLMs.
  arXiv  Detail & Related papers  (2025-02-26T08:07:17Z)
• CQIL: Inference Latency Optimization with Concurrent Computation of Quasi-Independent Layers [21.91815582658188]
  Large scale language models are delivering unprecedented performance on almost all natural language processing tasks. The overwhelming complexity incurs a high inference latency that negatively affects user experience. We propose to identify quasi-independent layers, which can be concurrently computed to significantly decrease inference latency.
  arXiv  Detail & Related papers  (2024-04-10T03:30:01Z)
• FFN-SkipLLM: A Hidden Gem for Autoregressive Decoding with Adaptive Feed Forward Skipping [49.66872823080736]
  Autoregressive Large Language Models (e.g., LLaMa, GPTs) are omnipresent achieving remarkable success in language understanding and generation. To mitigate overload incurred during generation, several early-exit and layer-dropping strategies have been proposed. We propose FFN-SkipLLM, which is an input-adaptive feed-forward skipping strategy.
  arXiv  Detail & Related papers  (2024-04-05T02:35:43Z)

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.