Zhuang Liu's Homepage

Research Scientist

Meta AI Research

Email: liuzhuangthu at gmail.com

Google Scholar / GitHub /
LinkedIn / Twitter

Zhuang Liu

I'm a Research Scientist at Meta AI Research. I received my PhD in Computer Science from UC Berkeley, advised by Prof. Trevor Darrell. I was also fortunate to have worked as a visiting researcher or an intern at Cornell University, Intel Labs, Adobe Research and FAIR. I earned my Bachelor's degree in Computer Science from Yao Class at Tsinghua University.

My research focuses on scaling neural networks both up and down, for building capable models and understanding their behaviors under different compute and data regimes. I place an emphasis on exploring simple and baseline approaches to gain insights into the workings of deep learning. My research is often characterized by challenging existing beliefs in the field .

I led the development of DenseNet (which won a CVPR Best Paper Award) and ConvNeXt. Both are among the most widely used neural architectures in computer vision and deep learning.

Selected Publications (* equal contribution)

A Simple and Effective Pruning Approach for Large Language Models

Mingjie Sun*, Zhuang Liu*, Anna Bair, Zico Kolter

arXiv 2023

[Paper] [Code]

Dropout Reduces Underfitting

Zhuang Liu*, Zhiqiu Xu*, Joseph Jin, Zhiqiang Shen, Trevor Darrell

ICML 2023

[Paper] [Code]

A ConvNet for the 2020s

Zhuang Liu, Hanzi Mao, Chao-Yuan Wu, Christoph Feichtenhofer, Trevor Darrell, Saining Xie

CVPR 2022

[Paper] [Video] [Code]

Rethinking the Value of Network Pruning

Zhuang Liu*, Mingjie Sun*, Tinghui Zhou, Gao Huang, Trevor Darrell

ICLR 2019

[Paper] [OpenReview] [Code]

NeurIPS'18 Compact Neural Networks Workshop Best Paper Award

Learning Efficient Convolutional Networks through Network Slimming

Zhuang Liu, Jianguo Li, Zhiqiang Shen, Gao Huang, Shoumeng Yan, Changshui Zhang

ICCV 2017

[Paper] [Code1] [Code2] [Code3 (3rd-party)]

Densely Connected Convolutional Networks

Gao Huang*, Zhuang Liu*, Laurens van der Maaten, Kilian Weinberger

CVPR 2017

[Paper] [Code]

CVPR Best Paper Award

Research Scientist

Meta AI Research

Email: liuzhuangthu at gmail.com

Google Scholar / GitHub /
LinkedIn / Twitter

Zhuang Liu

Selected Publications (* equal contribution)

A Simple and Effective Pruning Approach for Large Language Models

Dropout Reduces Underfitting

A ConvNet for the 2020s

Rethinking the Value of Network Pruning

Learning Efficient Convolutional Networks through Network Slimming

Densely Connected Convolutional Networks

All Publications (* equal contribution)

A Simple and Effective Pruning Approach for Large Language Models

One-for-All: Generalized LoRA for Parameter-Efficient Fine-tuning

ImageBind: One Embedding Space To Bind Them All

Dropout Reduces Underfitting

ConvNeXt V2: Co-designing and Scaling ConvNets with Masked Autoencoders

Exploring Simple and Transferable Recognition-Aware Image Processing

A ConvNet for the 2020s

Anytime Dense Prediction with Confidence Adaptivity

Vision Transformer Slimming: Multi-Dimension Searching in Continuous Optimization Space

Un-Mix: Rethinking Image Mixtures for Unsupervised Visual Representation Learning

Meta-Baseline: Exploring Simple Meta-Learning for Few-Shot Learning

SensAI: Fast ConvNets Serving on Live Data via Class Parallelism

Regularization Matters in Policy Optimization - An Empirical Study on Continuous Control

Test-Time Training for Out-of-Distribution Generalization

MSeg: A Composite Dataset for Multi-domain Semantic Segmentation

Few Sample Knowledge Distillation for Efficient Network Compression

Few-shot Object Detection via Feature Reweighting

Rethinking the Value of Network Pruning

DSOD: Learning Deeply Supervised Object Detectors from Scratch

Learning Efficient Convolutional Networks through Network Slimming

Densely Connected Convolutional Networks

Snapshot Ensembles: Train 1, Get M for Free

Deep Networks with Stochastic Depth

Misc

Research Scientist

Meta AI Research

Email: liuzhuangthu at gmail.com

Google Scholar / GitHub / LinkedIn / Twitter

Zhuang Liu

Selected Publications (* equal contribution)

A Simple and Effective Pruning Approach for Large Language Models

Dropout Reduces Underfitting

A ConvNet for the 2020s

Rethinking the Value of Network Pruning

Learning Efficient Convolutional Networks through Network Slimming

Densely Connected Convolutional Networks

All Publications (* equal contribution)

A Simple and Effective Pruning Approach for Large Language Models

One-for-All: Generalized LoRA for Parameter-Efficient Fine-tuning

ImageBind: One Embedding Space To Bind Them All

Dropout Reduces Underfitting

ConvNeXt V2: Co-designing and Scaling ConvNets with Masked Autoencoders

Exploring Simple and Transferable Recognition-Aware Image Processing

A ConvNet for the 2020s

Anytime Dense Prediction with Confidence Adaptivity

Vision Transformer Slimming: Multi-Dimension Searching in Continuous Optimization Space

Un-Mix: Rethinking Image Mixtures for Unsupervised Visual Representation Learning

Meta-Baseline: Exploring Simple Meta-Learning for Few-Shot Learning

SensAI: Fast ConvNets Serving on Live Data via Class Parallelism

Regularization Matters in Policy Optimization - An Empirical Study on Continuous Control

Test-Time Training for Out-of-Distribution Generalization

MSeg: A Composite Dataset for Multi-domain Semantic Segmentation

Few Sample Knowledge Distillation for Efficient Network Compression

Few-shot Object Detection via Feature Reweighting

Rethinking the Value of Network Pruning

DSOD: Learning Deeply Supervised Object Detectors from Scratch

Learning Efficient Convolutional Networks through Network Slimming

Densely Connected Convolutional Networks

Snapshot Ensembles: Train 1, Get M for Free

Deep Networks with Stochastic Depth

Misc

Google Scholar / GitHub /
LinkedIn / Twitter