https://github.com/human-analysis/neural-architecture-transfer

Neural Architecture Transfer (Arxiv'20), PyTorch Implementation
https://github.com/human-analysis/neural-architecture-transfer

Last synced: about 1 month ago
JSON representation

Neural Architecture Transfer (Arxiv'20), PyTorch Implementation

• Host: GitHub
• URL: https://github.com/human-analysis/neural-architecture-transfer
• Owner: human-analysis
• Created: 2020-05-01T01:31:52.000Z (over 4 years ago)
• Default Branch: master
• Last Pushed: 2020-05-15T18:42:30.000Z (over 4 years ago)
• Last Synced: 2024-08-03T09:03:23.063Z (5 months ago)
• Language: Python
• Homepage: http://hal.cse.msu.edu/papers/neural-architecture-transfer/
• Size: 4.18 MB
• Stars: 150
• Watchers: 12
• Forks: 21
• Open Issues: 6
• Metadata Files:
  • Readme: README.md

Awesome Lists containing this project

• StarryDivineSky - human-analysis/neural-architecture-transfer

README

        
# Neural Architecture Transfer

[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/neural-architecture-transfer/neural-architecture-search-on-imagenet)](https://paperswithcode.com/sota/neural-architecture-search-on-imagenet?p=neural-architecture-transfer)


[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/neural-architecture-transfer/architecture-search-on-cifar-10-image)](https://paperswithcode.com/sota/architecture-search-on-cifar-10-image?p=neural-architecture-transfer)


[![PWC](https://img.shields.io/endpoint.svg?url=https://paperswithcode.com/badge/neural-architecture-transfer/neural-architecture-search-on-cifar-100-1)](https://paperswithcode.com/sota/neural-architecture-search-on-cifar-100-1?p=neural-architecture-transfer)


Code accompanying the paper. 

> [Neural Architecture Transfer](https://arxiv.org/abs/2005.05859)

>

> [Zhichao Lu](https://www.zhichaolu.com), [Gautam Sreekumar](http://hal.cse.msu.edu/team/gautam-sreekumar/), [Erik Goodman](https://www.egr.msu.edu/~goodman/), [Wolfgang Banzhaf](http://www.cse.msu.edu/~banzhafw/), [Kalyanmoy Deb](https://www.egr.msu.edu/~kdeb/), and [Vishnu Boddeti](http://hal.cse.msu.edu/team/vishnu-boddeti/)

>

> *arXiv:2005.05859*

![overview](assets/images/overview.png)

## Requirements

``` 

Python >= 3.7.x, PyTorch >= 1.4.0, timm >= 0.1.18 

```

## ImageNet Classification

![imagenet](assets/images/imagenet.png)

``` shell

python evaluator.py --data /path/to/dataset --model subnets/imagenet/NAT-{M1,M2,M3,M4}/net.config

```

## Architecture Transfer

![transfer](assets/images/dataset.png)

``` shell

python evaluator.py \

  --data /path/to/dataset \

  --dataset {aircraft,cars,cifar10,cifar100,cinic10,dtd,flowers102,food101,pets,stl10} \

  --model subnets/{dataset}/net-img@{xxx}-flops@{xxx}-top1@{xx.x}/net.config

```

## Scalability to Objectives

#### Tri-Objective

![tri-obj](assets/images/tri_obj.png)

#### 12-Objective

![12-obj](assets/images/12_obj.png)

#### Architecture Visualization

![visualization](assets/images/archs.png)

## Acknowledgement 

Codes are modified from [OnceForAll](https://github.com/mit-han-lab/once-for-all) and [pytorch-image-models](https://github.com/rwightman/pytorch-image-models)