Up to 36x Speedup: Mask-based Parallel Inference Paradigm for Key Information Extraction in MLLMs

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Wang, Xinzhong, Guo, Ya, Li, Jing, Chen, Huan, Tu, Yi, Hong, Yijie, Liu, Gongshen, Zhu, Huijia
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866917225293676544
author Wang, Xinzhong
Guo, Ya
Li, Jing
Chen, Huan
Tu, Yi
Hong, Yijie
Liu, Gongshen
Zhu, Huijia
author_facet Wang, Xinzhong
Guo, Ya
Li, Jing
Chen, Huan
Tu, Yi
Hong, Yijie
Liu, Gongshen
Zhu, Huijia
contents Key Information Extraction (KIE) from visually-rich documents (VrDs) is a critical task, for which recent Large Language Models (LLMs) and Multi-Modal Large Language Models (MLLMs) have demonstrated strong potential. However, their reliance on autoregressive inference, which generates outputs sequentially, creates a significant efficiency bottleneck, especially as KIE tasks often involve extracting multiple, semantically independent fields. To overcome this limitation, we introduce PIP: a Parallel Inference Paradigm for KIE. Our approach reformulates the problem by using "[mask]" tokens as placeholders for all target values, enabling their simultaneous generation in a single forward pass. To facilitate this paradigm, we develop a tailored mask pre-training strategy and construct large-scale supervised datasets. Experimental results show that our PIP-models achieve a 5-36x inference speedup with negligible performance degradation compared to traditional autoregressive base models. By substantially improving efficiency while maintaining high accuracy, PIP paves the way for scalable and practical real-world KIE solutions.
format Preprint
id arxiv_https___arxiv_org_abs_2601_19613
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Up to 36x Speedup: Mask-based Parallel Inference Paradigm for Key Information Extraction in MLLMs
Wang, Xinzhong
Guo, Ya
Li, Jing
Chen, Huan
Tu, Yi
Hong, Yijie
Liu, Gongshen
Zhu, Huijia
Computation and Language
Artificial Intelligence
Key Information Extraction (KIE) from visually-rich documents (VrDs) is a critical task, for which recent Large Language Models (LLMs) and Multi-Modal Large Language Models (MLLMs) have demonstrated strong potential. However, their reliance on autoregressive inference, which generates outputs sequentially, creates a significant efficiency bottleneck, especially as KIE tasks often involve extracting multiple, semantically independent fields. To overcome this limitation, we introduce PIP: a Parallel Inference Paradigm for KIE. Our approach reformulates the problem by using "[mask]" tokens as placeholders for all target values, enabling their simultaneous generation in a single forward pass. To facilitate this paradigm, we develop a tailored mask pre-training strategy and construct large-scale supervised datasets. Experimental results show that our PIP-models achieve a 5-36x inference speedup with negligible performance degradation compared to traditional autoregressive base models. By substantially improving efficiency while maintaining high accuracy, PIP paves the way for scalable and practical real-world KIE solutions.
title Up to 36x Speedup: Mask-based Parallel Inference Paradigm for Key Information Extraction in MLLMs
topic Computation and Language
Artificial Intelligence
url https://arxiv.org/abs/2601.19613