Zobrazeno 1 - 10
of 1 924
pro vyhledávání: '"Han Xiaodong"'
Autor:
Yang, Dongjie, Huang, Suyuan, Lu, Chengqiang, Han, Xiaodong, Zhang, Haoxin, Gao, Yan, Hu, Yao, Zhao, Hai
Advancements in multimodal learning, particularly in video understanding and generation, require high-quality video-text datasets for improved model performance. Vript addresses this issue with a meticulously annotated corpus of 12K high-resolution v
Externí odkaz:
http://arxiv.org/abs/2406.06040
Large Language Models (LLMs) have shown remarkable comprehension abilities but face challenges in GPU memory usage during inference, hindering their scalability for real-time applications like chatbots. To accelerate inference, we store computed keys
Externí odkaz:
http://arxiv.org/abs/2405.12532
Autor:
Yang, Luyan, Savchenko, Andrii S., Zheng, Fengshan, Kiselev, Nikolai S., Rybakov, Filipp N., Han, Xiaodong, Blügel, Stefan, Dunin-Borkowski, Rafal E.
Publikováno v:
Advanced Materials 2024, 2403274
Magnetic skyrmions are topologically nontrivial spin configurations that possess particle-like properties. Earlier research was mainly focused on a specific type of skyrmion with topological charge Q = -1. However, theoretical analyses of two-dimensi
Externí odkaz:
http://arxiv.org/abs/2403.16931
Autor:
Wang, Xiangtong, Han, Xiaodong, Yang, Menglong, Xing, Chuan, Wang, Yuqi, Han, Songchen, Li, Wei
Low Earth orbit (LEO) mega-constellations rely on inter-satellite links (ISLs) to provide global connectivity. We note that in addition to the general constellation parameters, the ISL spanning patterns are also greatly influence the final network st
Externí odkaz:
http://arxiv.org/abs/2312.15873
The aim of audio-visual segmentation (AVS) is to precisely differentiate audible objects within videos down to the pixel level. Traditional approaches often tackle this challenge by combining information from various modalities, where the contributio
Externí odkaz:
http://arxiv.org/abs/2308.08288
Autor:
Qin, Zhen, Li, Dong, Sun, Weigao, Sun, Weixuan, Shen, Xuyang, Han, Xiaodong, Wei, Yunshen, Lv, Baohong, Luo, Xiao, Qiao, Yu, Zhong, Yiran
We present TransNormerLLM, the first linear attention-based Large Language Model (LLM) that outperforms conventional softmax attention-based models in terms of both accuracy and efficiency. TransNormerLLM evolves from the previous linear attention ar
Externí odkaz:
http://arxiv.org/abs/2307.14995
Autor:
Qin, Zhen, Sun, Weixuan, Lu, Kaiyue, Deng, Hui, Li, Dongxu, Han, Xiaodong, Dai, Yuchao, Kong, Lingpeng, Zhong, Yiran
Relative positional encoding is widely used in vanilla and linear transformers to represent positional information. However, existing encoding methods of a vanilla transformer are not always directly applicable to a linear transformer, because the la
Externí odkaz:
http://arxiv.org/abs/2307.09270
Autor:
Qin, Zhen, Han, Xiaodong, Sun, Weixuan, He, Bowen, Li, Dong, Li, Dongxu, Dai, Yuchao, Kong, Lingpeng, Zhong, Yiran
Sequence modeling has important applications in natural language processing and computer vision. Recently, the transformer-based models have shown strong performance on various sequence modeling tasks, which rely on attention to capture pairwise toke
Externí odkaz:
http://arxiv.org/abs/2305.04749
Autor:
Shen, Xuyang, Li, Dong, Zhou, Jinxing, Qin, Zhen, He, Bowen, Han, Xiaodong, Li, Aixuan, Dai, Yuchao, Kong, Lingpeng, Wang, Meng, Qiao, Yu, Zhong, Yiran
We explore a new task for audio-visual-language modeling called fine-grained audible video description (FAVD). It aims to provide detailed textual descriptions for the given audible videos, including the appearance and spatial locations of each objec
Externí odkaz:
http://arxiv.org/abs/2303.15616
Autor:
Qin, Zhen, Han, XiaoDong, Sun, Weixuan, Li, Dongxu, Kong, Lingpeng, Barnes, Nick, Zhong, Yiran
Linear transformers aim to reduce the quadratic space-time complexity of vanilla transformers. However, they usually suffer from degraded performances on various tasks and corpus. In this paper, we examine existing kernel-based linear transformers an
Externí odkaz:
http://arxiv.org/abs/2210.10340