EnsemJudge: Enhancing Reliability in Chinese LLM-Generated Text Detection through Diverse Model Ensembles

TL;DR

EnsemJudge is a robust ensemble-based framework designed to improve the detection of Chinese texts generated by large language models, outperforming baselines and winning a shared task.

Contribution

It introduces a novel ensemble voting mechanism tailored for Chinese LLM detection, addressing out-of-domain and adversarial challenges.

Findings

Outperformed all baseline methods in Chinese LLM detection

Achieved first place in NLPCC2025 Shared Task 1

Demonstrated robustness against out-of-domain and adversarial samples

Abstract

Large Language Models (LLMs) are widely applied across various domains due to their powerful text generation capabilities. While LLM-generated texts often resemble human-written ones, their misuse can lead to significant societal risks. Detecting such texts is an essential technique for mitigating LLM misuse, and many detection methods have shown promising results across different datasets. However, real-world scenarios often involve out-of-domain inputs or adversarial samples, which can affect the performance of detection methods to varying degrees. Furthermore, most existing research has focused on English texts, with limited work addressing Chinese text detection. In this study, we propose EnsemJudge, a robust framework for detecting Chinese LLM-generated text by incorporating tailored strategies and ensemble voting mechanisms. We trained and evaluated our system on a carefully constructed Chinese dataset provided by NLPCC2025 Shared Task 1. Our approach outperformed all baseline methods and achieved first place in the task, demonstrating its effectiveness and reliability in Chinese LLM-generated text detection. Our code is available at https://github.com/johnsonwangzs/MGT-Mini.