Výsledky vyhledávání - "Jiang, LiJuan"

Report

Efficient Training of Large Language Models on Distributed Infrastructures: A Survey

Autor: Duan, Jiangfei, Zhang, Shuo, Wang, Zerui, Jiang, Lijuan, Qu, Wenwen, Hu, Qinghao, Wang, Guoteng, Weng, Qizhen, Yan, Hang, Zhang, Xingcheng, Qiu, Xipeng, Lin, Dahua, Wen, Yonggang, Jin, Xin, Zhang, Tianwei, Sun, Peng

Large Language Models (LLMs) like GPT and LLaMA are revolutionizing the AI industry with their sophisticated capabilities. Training these models requires vast GPU clusters and significant computing time, posing major challenges in terms of scalabilit

Externí odkaz: http://arxiv.org/abs/2407.20018

Zobrazit plný text záznamu

Report

ZeroPP: Unleashing Exceptional Parallelism Efficiency through Tensor-Parallelism-Free Methodology

Autor: Tang, Ding, Jiang, Lijuan, Zhou, Jiecheng, Jin, Minxi, Li, Hengjie, Zhang, Xingcheng, Pei, Zhilin, Zhai, Jidong

Large-scale models rely heavily on 3D parallelism for distributed training, which utilizes tensor parallelism (TP) as the intra-operator parallelism to partition model states across GPUs. However, TP introduces significant communication overheads and

Externí odkaz: http://arxiv.org/abs/2402.03791

Zobrazit plný text záznamu

Akademický článek

Cryoablation techniques in bladder cancer: A review

Autor: Ma Binglei, Teixeira Wilhem, Jiang Lijuan

Publikováno v: Frigid Zone Medicine, Vol 4, Iss 2, Pp 72-77 (2024)

Bladder cancer (BC) ranks as the tenth most common cancer globally. Histopathologically, BC is broadly categorized into urothelial and non-urothelial BC. Urothelial carcinoma represents over 90% of BC in most regions worldwide. The standard treatment

Externí odkaz: https://doaj.org/article/19209d9163834a18b17cc6352ac5fa4f