Building a Taiwanese Mandarin Spoken Language Model: A First Attempt
Authors:
Chih-Kai Yang,
Yu-Kuan Fu,
Chen-An Li,
Yi-Cheng Lin,
Yu-Xiang Lin,
Wei-Chih Chen,
Ho Lam Chung,
Chun-Yi Kuan,
Wei-Ping Huang,
Ke-Han Lu,
Tzu-Quan Lin,
Hsiu-Hsuan Wang,
En-Pei Hu,
Chan-Jan Hsu,
Liang-Hsuan Tseng,
I-Hsiang Chiu,
Ulin Sanga,
Xuanjun Chen,
Po-chun Hsu,
Shu-wen Yang,
Hung-yi Lee
Abstract:
This technical report presents our initial attempt to build a spoken large language model (LLM) for Taiwanese Mandarin, specifically tailored to enable real-time, speech-to-speech interaction in multi-turn conversations. Our end-to-end model incorporates a decoder-only transformer architecture and aims to achieve seamless interaction while preserving the conversational flow, including full-duplex…
▽ More
This technical report presents our initial attempt to build a spoken large language model (LLM) for Taiwanese Mandarin, specifically tailored to enable real-time, speech-to-speech interaction in multi-turn conversations. Our end-to-end model incorporates a decoder-only transformer architecture and aims to achieve seamless interaction while preserving the conversational flow, including full-duplex capabilities allowing simultaneous speaking and listening. The paper also details the training process, including data preparation with synthesized dialogues and adjustments for real-time interaction. We also developed a platform to evaluate conversational fluency and response coherence in multi-turn dialogues. We hope the release of the report can contribute to the future development of spoken LLMs in Taiwanese Mandarin.
△ Less
Submitted 27 December, 2024; v1 submitted 11 November, 2024;
originally announced November 2024.
Navigating simplicity and complexity of social-ecological systems through a dialog between dynamical systems and agent-based models
Authors:
Sonja Radosavljevic,
Udita Sanga,
Maja Schlüter
Abstract:
Social-ecological systems research aims to understand the nature of social-ecological phenomena, to find ways to foster or manage conditions under which desired phenomena occur or to reduce the negative consequences of undesirable phenomena. Such challenges are often addressed using dynamical systems models (DSM) or agent-based models (ABM). Here we develop an iterative procedure for combining DSM…
▽ More
Social-ecological systems research aims to understand the nature of social-ecological phenomena, to find ways to foster or manage conditions under which desired phenomena occur or to reduce the negative consequences of undesirable phenomena. Such challenges are often addressed using dynamical systems models (DSM) or agent-based models (ABM). Here we develop an iterative procedure for combining DSM and ABM to leverage their strengths and gain insights that surpass insights obtained by each approach separately. The procedure uses results of an ABM as inputs for a DSM development. In the following steps, results of the DSM analyses guide future analysis of the ABM and vice versa. This dialogue, more than having a tight connection between the models, enables pushing the research frontier, expanding the set of research questions and insights. We illustrate our method with the example of poverty traps and innovation in agricultural systems, but our conclusions are general and can be applied to other DSM-ABM combinations.
△ Less
Submitted 24 June, 2024; v1 submitted 19 February, 2024;
originally announced February 2024.
