arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2402.04838cs.CLcs.AI
收藏

PaDeLLM-NER:面向命名实体识别的大语言模型并行解码

PaDeLLM-NER: Parallel Decoding in Large Language Models for Named Entity Recognition

ByteDance(字节跳动);University of Chinese Academy of Sciences(中国科学院大学);University College Dublin(都柏林大学学院)

Jinghui Lu, Ziwei Yang, Yanjie Wang, Xuejing Liu, Brian Mac Namee, Can Huang

AI总结:

针对大语言模型在命名实体识别中因顺序解码导致的高延迟问题,本文提出无需额外模块的并行解码方法 PaDeLLM-NER,在保持预测质量的同时将推理速度提升 1.76 至 10.22 倍。

摘要:

在本研究中,我们旨在降低使用大语言模型(LLMs)进行命名实体识别(NER)时的生成延迟。LLMs 高延迟的主要原因是顺序解码过程,该过程以自回归方式生成 NER 的所有标签和提及,显著增加了序列长度。为此,我们提出了面向 NER 的 LLM 并行解码方法 PaDeLLM-NER,该方法能够无缝集成到现有生成式模型框架中,无需额外模块或架构修改。PaDeLLM-NER 允许同时解码所有提及,从而降低生成延迟。实验表明,PaDeLLM-NER 显著提高了推理速度,在英文和中文上比自回归方法快 1.76 到 10.22 倍。同时,它保持了预测质量,其在多个数据集上的性能与最先进方法相当。

英文摘要:

In this study, we aim to reduce generation latency for Named Entity Recognition (NER) with Large Language Models (LLMs). The main cause of high latency in LLMs is the sequential decoding process, which autoregressively generates all labels and mentions for NER, significantly increase the sequence length. To this end, we introduce Parallel Decoding in LLM for NE} (PaDeLLM-NER), a approach that integrates seamlessly into existing generative model frameworks without necessitating additional modules or architectural modifications. PaDeLLM-NER allows for the simultaneous decoding of all mentions, thereby reducing generation latency. Experiments reveal that PaDeLLM-NER significantly increases inference speed that is 1.76 to 10.22 times faster than the autoregressive approach for both English and Chinese. Simultaneously it maintains the quality of predictions as evidenced by the performance that is on par with the state-of-the-art across various datasets.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑