Zobrazeno 1 - 10
of 1 159
pro vyhledávání: '"Zhang, XinRong"'
One essential advantage of recurrent neural networks (RNNs) over transformer-based language models is their linear computational complexity concerning the sequence length, which makes them much faster in handling long sequences during inference. Howe
Externí odkaz:
http://arxiv.org/abs/2410.07145
Autor:
Zhang, Xinrong, Chen, Yingfa, Hu, Shengding, Han, Xu, Xu, Zihang, Xu, Yuanwei, Zhao, Weilin, Sun, Maosong, Liu, Zhiyuan
As large language models (LLMs) increasingly permeate daily lives, there is a growing demand for real-time interactions that mirror human conversations. Traditional turn-based chat systems driven by LLMs prevent users from verbally interacting with t
Externí odkaz:
http://arxiv.org/abs/2406.15718
Autor:
Sun, Ao, Zhao, Weilin, Han, Xu, Yang, Cheng, Zhang, Xinrong, Liu, Zhiyuan, Shi, Chuan, Sun, Maosong
The emergence of large language models (LLMs) relies heavily on distributed training strategies, among which pipeline parallelism plays a crucial role. As LLMs' training sequence length extends to 32k or even 128k, the current pipeline parallel metho
Externí odkaz:
http://arxiv.org/abs/2406.03488
Autor:
Hu, Shengding, Tu, Yuge, Han, Xu, He, Chaoqun, Cui, Ganqu, Long, Xiang, Zheng, Zhi, Fang, Yewei, Huang, Yuxiang, Zhao, Weilin, Zhang, Xinrong, Thai, Zheng Leng, Zhang, Kaihuo, Wang, Chongyi, Yao, Yuan, Zhao, Chenyang, Zhou, Jie, Cai, Jie, Zhai, Zhongwu, Ding, Ning, Jia, Chao, Zeng, Guoyang, Li, Dahai, Liu, Zhiyuan, Sun, Maosong
The burgeoning interest in developing Large Language Models (LLMs) with up to trillion parameters has been met with concerns regarding resource efficiency and practical expense, particularly given the immense cost of experimentation. This scenario un
Externí odkaz:
http://arxiv.org/abs/2404.06395
Autor:
Zhao, Weilin, Huang, Yuxiang, Han, Xu, Xu, Wang, Xiao, Chaojun, Zhang, Xinrong, Fang, Yewei, Zhang, Kaihuo, Liu, Zhiyuan, Sun, Maosong
Speculative decoding is a widely used method that accelerates the generation process of large language models (LLMs) with no compromise in model performance. It achieves this goal by using an existing smaller model for drafting and then employing the
Externí odkaz:
http://arxiv.org/abs/2402.13720
Autor:
Zhang, Xinrong, Chen, Yingfa, Hu, Shengding, Xu, Zihang, Chen, Junhao, Hao, Moo Khai, Han, Xu, Thai, Zhen Leng, Wang, Shuo, Liu, Zhiyuan, Sun, Maosong
Publikováno v:
2023.12.15ARR
Processing and reasoning over long contexts is crucial for many practical applications of Large Language Models (LLMs), such as document comprehension and agent construction. Despite recent strides in making LLMs process contexts with more than 100K
Externí odkaz:
http://arxiv.org/abs/2402.13718
Autor:
Hu, Shengding, Liu, Xin, Han, Xu, Zhang, Xinrong, He, Chaoqun, Zhao, Weilin, Lin, Yankai, Ding, Ning, Ou, Zebin, Zeng, Guoyang, Liu, Zhiyuan, Sun, Maosong
The scientific scale-up of large language models (LLMs) necessitates a comprehensive understanding of their scaling properties. However, the existing literature on the scaling properties only yields an incomplete answer: optimization loss decreases p
Externí odkaz:
http://arxiv.org/abs/2310.03262
Autor:
He, Rongrong, Lv, Ziwei, Li, Yinan, Ren, Shuchao, Cao, Jiaqi, Zhu, Jun, Zhang, Xinrong, Wu, Huimin, Wan, Lihao, Tang, Ji, Xu, Shutong, Chen, Xiao-Lin, Zhou, Zhipeng
Publikováno v:
In Developmental Cell 18 November 2024 59(22):2931-2946
Autor:
Zhang, Xinrong, Ren, Zihou, Li, Xi, Liu, Shuqi, Deng, Yunlong, Xiao, Yadi, Han, Yuxing, Wen, Jiangtao
The deluge of new papers has significantly blocked the development of academics, which is mainly caused by author-level and publication-level evaluation metrics that only focus on quantity. Those metrics have resulted in several severe problems that
Externí odkaz:
http://arxiv.org/abs/2201.02915
Autor:
Tu, You, Wang, Xiaofeng, Zhou, Jitao, Wang, Xiaoxue, Jia, Zixu, Ma, Jiahao, Yao, Wenjie, Zhang, Xinrong, Sun, Zechong, Luo, Pingping, Feng, Xiaoming, Fu, Bojie
Publikováno v:
In Agricultural and Forest Meteorology 15 August 2024 355