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

Chrome logo  Add to Chrome

Firefox logo Add to Firefox
Picture for Xiaodan Song

Xiaodan Song
Alert button

Token Dropping for Efficient BERT Pretraining


Mar 24, 2022
Le Hou, Richard Yuanzhe Pang, Tianyi Zhou, Yuexin Wu, Xinying Song, Xiaodan Song, Denny Zhou

Add code
Alert button

Figure 1 for Token Dropping for Efficient BERT Pretraining
Figure 2 for Token Dropping for Efficient BERT Pretraining
Figure 3 for Token Dropping for Efficient BERT Pretraining
Figure 4 for Token Dropping for Efficient BERT Pretraining
* ACL 2022 
   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

Auto-scaling Vision Transformers without Training


Feb 27, 2022
Wuyang Chen, Wei Huang, Xianzhi Du, Xiaodan Song, Zhangyang Wang, Denny Zhou

Add code
Alert button

Figure 1 for Auto-scaling Vision Transformers without Training
Figure 2 for Auto-scaling Vision Transformers without Training
Figure 3 for Auto-scaling Vision Transformers without Training
Figure 4 for Auto-scaling Vision Transformers without Training
* ICLR 2022 accepted 
   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

A Simple Single-Scale Vision Transformer for Object Localization and Instance Segmentation


Dec 17, 2021
Wuyang Chen, Xianzhi Du, Fan Yang, Lucas Beyer, Xiaohua Zhai, Tsung-Yi Lin, Huizhong Chen, Jing Li, Xiaodan Song, Zhangyang Wang, Denny Zhou

Add code
Alert button

Figure 1 for A Simple Single-Scale Vision Transformer for Object Localization and Instance Segmentation
Figure 2 for A Simple Single-Scale Vision Transformer for Object Localization and Instance Segmentation
Figure 3 for A Simple Single-Scale Vision Transformer for Object Localization and Instance Segmentation
Figure 4 for A Simple Single-Scale Vision Transformer for Object Localization and Instance Segmentation

   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

Speeding up Deep Model Training by Sharing Weights and Then Unsharing


Oct 08, 2021
Shuo Yang, Le Hou, Xiaodan Song, Qiang Liu, Denny Zhou

Add code
Alert button

Figure 1 for Speeding up Deep Model Training by Sharing Weights and Then Unsharing
Figure 2 for Speeding up Deep Model Training by Sharing Weights and Then Unsharing
Figure 3 for Speeding up Deep Model Training by Sharing Weights and Then Unsharing
Figure 4 for Speeding up Deep Model Training by Sharing Weights and Then Unsharing

   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

Efficient Scale-Permuted Backbone with Learned Resource Distribution


Oct 22, 2020
Xianzhi Du, Tsung-Yi Lin, Pengchong Jin, Yin Cui, Mingxing Tan, Quoc Le, Xiaodan Song

Add code
Alert button

Figure 1 for Efficient Scale-Permuted Backbone with Learned Resource Distribution
Figure 2 for Efficient Scale-Permuted Backbone with Learned Resource Distribution
Figure 3 for Efficient Scale-Permuted Backbone with Learned Resource Distribution
Figure 4 for Efficient Scale-Permuted Backbone with Learned Resource Distribution
* ECCV2020 
   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

Go Wide, Then Narrow: Efficient Training of Deep Thin Networks


Jul 01, 2020
Denny Zhou, Mao Ye, Chen Chen, Tianjian Meng, Mingxing Tan, Xiaodan Song, Quoc Le, Qiang Liu, Dale Schuurmans

Add code
Alert button

Figure 1 for Go Wide, Then Narrow: Efficient Training of Deep Thin Networks
Figure 2 for Go Wide, Then Narrow: Efficient Training of Deep Thin Networks
Figure 3 for Go Wide, Then Narrow: Efficient Training of Deep Thin Networks
Figure 4 for Go Wide, Then Narrow: Efficient Training of Deep Thin Networks
* ICML 2020 
   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices


Apr 14, 2020
Zhiqing Sun, Hongkun Yu, Xiaodan Song, Renjie Liu, Yiming Yang, Denny Zhou

Add code
Alert button

Figure 1 for MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices
Figure 2 for MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices
Figure 3 for MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices
Figure 4 for MobileBERT: a Compact Task-Agnostic BERT for Resource-Limited Devices
* Accepted to ACL 2020 
   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

BigNAS: Scaling Up Neural Architecture Search with Big Single-Stage Models


Mar 24, 2020
Jiahui Yu, Pengchong Jin, Hanxiao Liu, Gabriel Bender, Pieter-Jan Kindermans, Mingxing Tan, Thomas Huang, Xiaodan Song, Ruoming Pang, Quoc Le

Add code
Alert button

Figure 1 for BigNAS: Scaling Up Neural Architecture Search with Big Single-Stage Models
Figure 2 for BigNAS: Scaling Up Neural Architecture Search with Big Single-Stage Models
Figure 3 for BigNAS: Scaling Up Neural Architecture Search with Big Single-Stage Models
Figure 4 for BigNAS: Scaling Up Neural Architecture Search with Big Single-Stage Models
* Technical report 
   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email

SpineNet: Learning Scale-Permuted Backbone for Recognition and Localization


Dec 10, 2019
Xianzhi Du, Tsung-Yi Lin, Pengchong Jin, Golnaz Ghiasi, Mingxing Tan, Yin Cui, Quoc V. Le, Xiaodan Song

Add code
Alert button

Figure 1 for SpineNet: Learning Scale-Permuted Backbone for Recognition and Localization
Figure 2 for SpineNet: Learning Scale-Permuted Backbone for Recognition and Localization
Figure 3 for SpineNet: Learning Scale-Permuted Backbone for Recognition and Localization
Figure 4 for SpineNet: Learning Scale-Permuted Backbone for Recognition and Localization

   Access Paper or Ask Questions
  • Share via Twitter
  • Share via Facebook
  • Share via LinkedIn
  • Share via Whatsapp
  • Share via Messenger
  • Share via Email
1
2
>>