Accès ouvert
2026
conference-paper
OpenAlex
Chao Jin, Ziheng Jiang, Zhihao Bai, Zheng Zhong et autres
We present MegaScale-MoE, a production system tailored for the efficient training of large-scale mixture-of-experts (MoE) models. MoE emerges as a promising architecture to scale large language models (LLMs) to unprecedented sizes, thereby enhancing model performance. However, existing MoE training systems experience a …
cn, cr
(code pays fourni par la source)
Accès ouvert
2026
review
OpenAlex
Yan Wei, Xiong Xiao, Xixi Wu, Cong Hua Xie et autres
Background With the increasing use of assisted reproductive technology (ART), more women with endometriosis are achieving pregnancy through ART. However, the impact of endometriosis on pregnancy and perinatal outcomes following ART remains controversial. This study aimed to clarify the association between endometriosis …
cn
(code pays fourni par la source)
2026
article
OpenAlex
Cong Hua Xie, Jiahui Yue, WY He, Changbin Yu
cn
(code pays fourni par la source)
Accès ouvert
2025
preprint
OpenAlex
Lizhang Chen, Jonathan Li, Kaizhao Liang, Bo Su et autres
We introduce Cautious Weight Decay (CWD), a one-line, optimizer-agnostic modification that applies weight decay only to parameter coordinates whose signs align with the optimizer update. Unlike standard decoupled decay, which implicitly optimizes a regularized or constrained objective, CWD preserves the original loss …
Accès ouvert
2025
article
OpenAlex
Cong Hua Xie, Kepeng Wang, Kai Yang, Yuanyuan Zhong et autres
Abstract Chronic obstructive pulmonary disease (COPD) is a heterogeneous disorder characterized by diverse clinical manifestations, pathophysiological mechanisms, and therapeutic responses. This review explores the evolving landscape of precision medicine in COPD management, with particular emphasis on optimizing patient care through the integration …
cn
(code pays fourni par la source)
Accès ouvert
2025
article
OpenAlex
Malcolm J. Borg, Cong Hua Xie, Chang Chen, Michelle J. Bound et autres
au, cn
(code pays fourni par la source)
2025
article
OpenAlex
Qiyun Wang, Jian Kang, Cong Hua Xie, Shengyang Ou et autres
Abstract The intellectualization and informatization of railway subgrade monitoring is the current development trend. The change of physical and mechanical properties of the subgrade is the basis of intelligent identification. Therefore, it is necessary to reveal the influence of rainfall infiltration on …
cn
(code pays fourni par la source)
2025
article
OpenAlex
Daniel Robert Quast, Cong Hua Xie, Michelle J. Bound, Jacqueline Grivell et autres
Individuals with type 2 diabetes are at high risk of postprandial decrease in blood pressure (BP; i.e., a reduction in systolic BP of ≥20 mmHg, termed postprandial hypotension [PPH]), which increases the risk of falls and mortality. This study evaluated the effects …
de, au
(code pays fourni par la source)
Accès ouvert
2025
other
OpenAlex
Daniel Robert Quast, Cong Hua Xie, Michelle J. Bound, Jacqueline Grivell et autres
Individuals with type 2 diabetes are at high risk of postprandial falls in blood pressure (BP) (i.e., a reduction in systolic BP of ³20mmHg, termed postprandial hypotension (PPH)), which increases the risk of falls and mortality. This study evaluated the effects of …
de, au
(code pays fourni par la source)
Accès ouvert
2025
other
OpenAlex
Daniel Robert Quast, Cong Hua Xie, Michelle J. Bound, Jacqueline Grivell et autres
Individuals with type 2 diabetes are at high risk of postprandial falls in blood pressure (BP) (i.e., a reduction in systolic BP of ³20mmHg, termed postprandial hypotension (PPH)), which increases the risk of falls and mortality. This study evaluated the effects of …
Accès ouvert
2024
article
OpenAlex
Cong Hua Xie, Peter E. Iroga, Michelle J. Bound, Jacqueline Grivell et autres
au
(code pays fourni par la source)
Accès ouvert
2024
preprint
OpenAlex
Yanyue Xie, Zhi Zhang, Ding Zhou, Cong Hua Xie et autres
Mixture-of-Experts (MoE) architectures face challenges such as high memory consumption and redundancy in experts. Pruning MoE can reduce network weights while maintaining model performance. Motivated by the recent observation of emergent large magnitude features in Large Language Models (LLM) and MoE routing …