Get our free extension to see links to code for papers anywhere online!

Add to Chrome

Add to Firefox

Picture for Junyang Lin

Junyang Lin

Prompt Tuning for Generative Multimodal Pretrained Models

Aug 04, 2022
Hao Yang, Junyang Lin, An Yang, Peng Wang, Chang Zhou, Hongxia Yang

Figure 1 for Prompt Tuning for Generative Multimodal Pretrained Models

Figure 2 for Prompt Tuning for Generative Multimodal Pretrained Models

Figure 3 for Prompt Tuning for Generative Multimodal Pretrained Models

Figure 4 for Prompt Tuning for Generative Multimodal Pretrained Models

* Work in progress

Access Paper or Ask Questions

Unifying Architectures, Tasks, and Modalities Through a Simple Sequence-to-Sequence Learning Framework

Feb 07, 2022
Peng Wang, An Yang, Rui Men, Junyang Lin, Shuai Bai, Zhikang Li, Jianxin Ma, Chang Zhou, Jingren Zhou, Hongxia Yang

Figure 1 for Unifying Architectures, Tasks, and Modalities Through a Simple Sequence-to-Sequence Learning Framework

Figure 2 for Unifying Architectures, Tasks, and Modalities Through a Simple Sequence-to-Sequence Learning Framework

Figure 3 for Unifying Architectures, Tasks, and Modalities Through a Simple Sequence-to-Sequence Learning Framework

Figure 4 for Unifying Architectures, Tasks, and Modalities Through a Simple Sequence-to-Sequence Learning Framework

* 23 pages, 11 figures

Access Paper or Ask Questions

KNAS: Green Neural Architecture Search

Nov 26, 2021
Jingjing Xu, Liang Zhao, Junyang Lin, Rundong Gao, Xu Sun, Hongxia Yang

Figure 1 for KNAS: Green Neural Architecture Search

Figure 2 for KNAS: Green Neural Architecture Search

Figure 3 for KNAS: Green Neural Architecture Search

Figure 4 for KNAS: Green Neural Architecture Search

* Accepted by ICML

Access Paper or Ask Questions

M6-10T: A Sharing-Delinking Paradigm for Efficient Multi-Trillion Parameter Pretraining

Oct 25, 2021
Junyang Lin, An Yang, Jinze Bai, Chang Zhou, Le Jiang, Xianyan Jia, Ang Wang, Jie Zhang, Yong Li, Wei Lin, Jingren Zhou, Hongxia Yang

Figure 1 for M6-10T: A Sharing-Delinking Paradigm for Efficient Multi-Trillion Parameter Pretraining

Figure 2 for M6-10T: A Sharing-Delinking Paradigm for Efficient Multi-Trillion Parameter Pretraining

Figure 3 for M6-10T: A Sharing-Delinking Paradigm for Efficient Multi-Trillion Parameter Pretraining

Figure 4 for M6-10T: A Sharing-Delinking Paradigm for Efficient Multi-Trillion Parameter Pretraining

* 14 pages, 4 figures

Access Paper or Ask Questions