AI Paper Atlas

Research map / Vision-language and multimodal / Multimodal LLMs

HSD: Training-Free Acceleration for Document Parsing Vision-Language Models with Hierarchical Speculative Decoding

Wenhui Liao, Hongliang Li, Pengyu Xie, Xinyu Cai, Yufan Shen, Yi Xin, Qi Qin, Shenglong Ye, Tianbin Li, Ming Hu, Junjun He, Yihao Liu and 6 more

ECCV 2026 · Vision-language and multimodal › Multimodal LLMs · approach: Training-free token reduction for MLLMs · 7 citations (OpenAlex)

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 changedWhereMeasuredEffect
add Stage 2 page-level verification after Stage 1 (ablation)HSD with dots.ocr · OmniDocBench v1.5accuracy: 70.47 -> 88.81Improved
add hierarchical two-stage design to pipeline drafter (ablation)HSD with dots.ocr · OmniDocBench v1.5end-to-end speedup: 2.09x -> 2.42xImproved
add hierarchical two-stage design to pipeline drafter (ablation)HSD with dots.ocr · OmniDocBench v1.5AAL: 2.49 -> 3.98Improved
use a fast pipeline drafter instead of autoregressive decoding (ablation)dots.ocr · OmniDocBench v1.5end-to-end speedup: 1.00x -> 2.09xImproved
add HSD to end-to-end parserHunyuanOCR · OmniDocBench v1.5mean latency (s/sample): 30.47 -> 10.96Improved
add HSD to end-to-end parserHunyuanOCR · OmniDocBench v1.5accuracy: 94.10 -> 94.02No clear effect
add HSD to end-to-end parserdots.ocr · OmniDocBench v1.5accuracy: 86.73 -> 88.81No clear effect
add HSD on top of visual token compressionDeepSeek-OCR · OmniDocBench v1.5end-to-end speedup: 1.00x -> 1.56xImproved
use HSD instead of ViSpec speculative decodingViSpec · OmniDocBench v1.5end-to-end speedup: 1.75x -> 2.12xImproved
lower acceptance threshold tau below 0.75 (ablation)HSD with dots.ocr · OmniDocBench v1.5parsing accuracyHurt
lower acceptance threshold tau (ablation)HSD with dots.ocr · OmniDocBench v1.5speedupImproved
degrade pipeline draft quality with injected noise (ablation)HSD with dots.ocr · OmniDocBench v1.5end-to-end speedupNo clear effect

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.

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

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.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy