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.
@article{arxiv.2409.11727,
title = {Enabling Real-Time Conversations with Minimal Training Costs},
author = {Wang Xu and Shuo Wang and Weilin Zhao and Xu Han and Yukun Yan and Yudi Zhang and Zhe Tao and Zhiyuan Liu and Wanxiang Che},
journal= {arXiv preprint arXiv:2409.11727},
year = {2024}
}