Chinese-Jev improves fast decision making for Chinese tasks
Chinese-Jev: Bringing System One Model to Chinese-Language Tasks
Computation and LanguageComputer Vision and Pattern Recognition
Summary
Some AI models work better when they just make decisions instead of writing long answers, but those for Chinese have been less accurate. The authors developed Chinese-Jev, a model trained on many examples and fine-tuned for medicine, law, and finance. It makes faster and more accurate decisions in Chinese than previous versions, and can even run efficiently on mobile devices. They also created a test set to measure its decision accuracy in different fields.
What this means in practice
- •For medical ai developers: Use Chinese-Jev to build faster medical decision tools that understand Chinese clinical questions more accurately.
- •For mobile app developers: Deploy Chinese-Jev on smartphones to enable quick, on-device decision making for Chinese language tasks.
Authors
Zexiao Wang, Zihao Zhang, Xudong Wang, Pan Wang, Ziyi Ye, Haoyu Zhao, Zuxuan Wu, Shuicheng Yan
Abstract
System One models such as Jev offer an efficient alternative to generative language models for tasks that require decisions rather than open-ended responses. However, existing Jev models exhibit limited Chinese-language decision accuracy, restricting their utility in both general and specialized settings. In this paper, we introduce Chinese-Jev, a System One model that addresses this gap through a unified data processing and training pipeline. Our data processing protocol converts heterogeneous Chinese-language annotations into probability targets over candidate options, enabling a shared training formulation across domains and question formats. To enable efficient inference, Chinese-Jev adopts a lightweight encoder-only backbone for text encoding and learns to score candidate answers through decision-oriented training. To address the misalignment between the pre-training distribution and downstream Chinese-language scenarios, we first train the model on a general-purpose corpus of 10 million examples, then fine-tune it separately for the medical, legal, and financial domains. To evaluate decision accuracy and calibration in both general and domain-specific Chinese-language settings, we introduce Chinese-Jev Bench (CJ-Bench). After first-stage pre-training, Chinese-Jev exceeds the accuracy of the closed-source Jev model by 1.24% on general-domain tasks while achieving a 20.3x speedup. Subsequent domain-specific fine-tuning yields a 4.0% accuracy improvement over Jev in medicine and achieves 92% of Jev's average accuracy across specialized domains, with a 17x speedup and an average latency of only 15 ms per example. We further demonstrate on-device deployment of an INT8-quantized model on mobile devices, achieving an inference latency of approximately 1.0 second per decision. The project is available at https://gulucaptain.github.io/Chinese-Jev/.