Enabling Real-Time Conversations with Minimal Training Costs

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Xu, Wang, Wang, Shuo, Zhao, Weilin, Han, Xu, Yan, Yukun, Zhang, Yudi, Tao, Zhe, Liu, Zhiyuan, Che, Wanxiang
Format: Preprint
Published: 2024
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866929503696060416
author Xu, Wang
Wang, Shuo
Zhao, Weilin
Han, Xu
Yan, Yukun
Zhang, Yudi
Tao, Zhe
Liu, Zhiyuan
Che, Wanxiang
author_facet Xu, Wang
Wang, Shuo
Zhao, Weilin
Han, Xu
Yan, Yukun
Zhang, Yudi
Tao, Zhe
Liu, Zhiyuan
Che, Wanxiang
contents Large language models (LLMs) have demonstrated the ability to improve human efficiency through conversational interactions. Conventional LLM-powered dialogue systems, operating on a turn-based paradigm, preclude real-time interaction during response generation. To address this limitation, researchers have proposed duplex models. These models can dynamically adapt to user input, facilitating real-time interactive feedback. However, these methods typically require substantial computational resources to acquire the ability. To reduce overhead, this paper presents a new duplex decoding approach that enhances LLMs with duplex ability, requiring minimal additional training. Specifically, our method employs parallel decoding of queries and responses in conversations, effectively implementing a channel-division-multiplexing decoding strategy. Experimental results indicate that our proposed method significantly enhances the naturalness and human-likeness of user-AI interactions with minimal training costs.
format Preprint
id arxiv_https___arxiv_org_abs_2409_11727
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle Enabling Real-Time Conversations with Minimal Training Costs
Xu, Wang
Wang, Shuo
Zhao, Weilin
Han, Xu
Yan, Yukun
Zhang, Yudi
Tao, Zhe
Liu, Zhiyuan
Che, Wanxiang
Computation and Language
Large language models (LLMs) have demonstrated the ability to improve human efficiency through conversational interactions. Conventional LLM-powered dialogue systems, operating on a turn-based paradigm, preclude real-time interaction during response generation. To address this limitation, researchers have proposed duplex models. These models can dynamically adapt to user input, facilitating real-time interactive feedback. However, these methods typically require substantial computational resources to acquire the ability. To reduce overhead, this paper presents a new duplex decoding approach that enhances LLMs with duplex ability, requiring minimal additional training. Specifically, our method employs parallel decoding of queries and responses in conversations, effectively implementing a channel-division-multiplexing decoding strategy. Experimental results indicate that our proposed method significantly enhances the naturalness and human-likeness of user-AI interactions with minimal training costs.
title Enabling Real-Time Conversations with Minimal Training Costs
topic Computation and Language
url https://arxiv.org/abs/2409.11727