Yi Xing

LLaDA MoE v2: Scaling Mixture-of-Experts Diffusion Language Models

Fengqi Zhu, Shaoxuan Xu, Jingyang Ou, Zebin You et autres

Diffusion language models (dLLMs) offer an alternative to autoregressive (AR) language modeling, yet the scaling behavior of Mixture-of-Experts (MoE) dLLMs remains poorly understood. We systematically characterize how optimization hyperparameters, compute allocation, and architecture scale for MoE dLLMs, identifying quantitative differences from scaling …

LLaDA MoE v2: Scaling Mixture-of-Experts Diffusion Language Models

Fengqi Zhu, Shaoxuan Xu, Jingyang Ou, Zebin You et autres

Diffusion language models (dLLMs) offer an alternative to autoregressive (AR) language modeling, yet the scaling behavior of Mixture-of-Experts (MoE) dLLMs remains poorly understood. We systematically characterize how optimization hyperparameters, compute allocation, and architecture scale for MoE dLLMs, identifying quantitative differences from scaling …

LLaDA2.1: Speeding Up Text Diffusion via Token Editing

Tiwei Bie, Maosong Cao, Xiang Cao, Bingsen Chen et autres

While LLaDA2.0 showcased the scaling potential of 100B-level block-diffusion models and their inherent parallelization, the delicate equilibrium between decoding speed and generation quality has remained an elusive frontier. Today, we unveil LLaDA2.1, a paradigm shift designed to transcend this trade-off. By seamlessly …

LLaDA2.1: Speeding Up Text Diffusion via Token Editing

Tiwei Bie, Maosong Cao, Xiang Cao, Bingsen Chen et autres

While LLaDA2.0 showcased the scaling potential of 100B-level block-diffusion models and their inherent parallelization, the delicate equilibrium between decoding speed and generation quality has remained an elusive frontier. Today, we unveil LLaDA2.1, a paradigm shift designed to transcend this trade-off. By seamlessly …

LLaDA2.0: Scaling Up Diffusion Language Models to 100B

Tiwei Bie, Kun Chen, Liang Du, Mengyan Gong et autres

This paper presents LLaDA2.0 -- a tuple of discrete diffusion large language models (dLLM) scaling up to 100B total parameters through systematic conversion from auto-regressive (AR) models -- establishing a new paradigm for frontier-scale deployment. Instead of costly training from scratch, LLaDA2.0 …

Wildfire-related PM2.5 and respiratory transmitted disease among Chinese children and adolescents from 2008 to 2019: A retrospective study

Li Chen, Rongbin Xu, Junqing Xie, Yi Xing et autres

BACKGROUND: Exposure to fine particles (PM2.5) from wildfires is known to cause deaths and chronic diseases, but its effect on respiratory infections, especially in children and adolescents, is not well characterized. We aimed to comprehensively assess the association between short-term exposure to …

fr, cn, au, gb (code pays fourni par la source)

LLaDA-MoE: A Sparse MoE Diffusion Language Model

Fengqi Zhu, Zebin You, Yi Xing, Zenan Huang et autres

We introduce LLaDA-MoE, a large language diffusion model with the Mixture-of-Experts (MoE) architecture, trained from scratch on approximately 20T tokens. LLaDA-MoE achieves competitive performance with significantly reduced computational overhead by maintaining a 7B-parameter capacity while activating only 1.4B parameters during inference. Our …

Impact of climate change and extreme temperature on the incidence of infectious disease among children and adolescents in China: A nationwide case-crossover study with over 8.7 million cases between 2008 and 2019

Li Chen, Duanke Liu, Yuchen Guo, Brian Wen et autres

BACKGROUND: The relationship between ambient temperature and infectious disease incidence lacks comprehensive documentation. Our study, therefore, sought to systematically determine the national association between temperature and the incidence of infectious diseases, categorized into respiratory, gastrointestinal and enterovirus, and vector-borne categories. We aimed …

cn, gb, au, nl (code pays fourni par la source)

Overview and Methods for Chinese National Surveillance on Students’ Common Diseases and Risk Factors, 2022

Yi Xing, Qi Ma, Mengjie Cui, La Mang et autres

Child and adolescent health plays a critical role in shaping future public health and intergenerational outcomes. In China, rising rates of myopia, obesity, mental health issues, and other common conditions highlight the need for continuous monitoring. Since 2016, the Chinese National Surveillance …

cn (code pays fourni par la source)

Strategic Co-Prevention Framework for Addressing Common Health Challenges Among Students in China

Yuechen Li, Yaqi Wang, Yihang Zhang, Jieyu Liu et autres

The concurrent rise in multiple health conditions among students, including myopia, obesity, spinal curvature abnormalities, and mental health disorders, represents a critical public health challenge. These interconnected health issues necessitate comprehensive prevention strategies that simultaneously address multiple conditions through synchronized behavioral and …

cn (code pays fourni par la source)

Tripartite Evolutionary Game Simulation Analysis of Carbon Emission Reduction Strategy from the Perspectives of Carbon Pricing and Taxation Mechanisms

Lu Feng, Fei Xaio-Ming, Jiang Zhong, Xiang Lei et autres

Under the dual-carbon goal and the carbon market mechanism framework, determining how the government, enterprises and emission reduction technology supplier groups strategically select emission reduction strategy by weighing respective benefits under different scenarios has emerged as a critical research question. Based on …

[Epidemiological characteristics and related factors of multimorbidity of common diseases among children and adolescents aged 7-18 years in Guangdong Province].

M Li, Songying Shen, Q X Chen, Rong Liu et autres

The multimorbidity rates of myopia and obesity, as well as myopia and malnutrition, in children and adolescents aged 7-18 in Guangdong Province are relatively low and fluctuate with age. Physical activity, screen time, consumption of sugary drinks, and sleep time may be …

cn (code pays fourni par la source)