Skip to content
Pytorch Implementation of: "Exploring Randomly Wired Neural Networks for Image Recognition"
Branch: master
Clone or download
Latest commit 8c729f6 Apr 16, 2019
Type Name Latest commit message Commit time
Failed to load latest commit information.
configs randwireNN Apr 5, 2019 Update Apr 16, 2019 update README Apr 6, 2019 randwireNN Apr 5, 2019 randwireNN Apr 5, 2019

RandWireNN(Randomly Wired Neural Network)

PyTorch implementation of : Exploring Randomly Wired Neural Networks for Image Recognition.


  • 2019/4/10: Release a result of regular computation(C=109) RandWird-WS(4,0.75). It has Top-1 accuracy of 77.07% on Imagenet dataset.
  • 2019/4/7: The code of RandWireNN are released.

Reproduced results

Model Paper's Top-1 Mine Top-1 Epochs LR Scheduler Weight Decay
RandWire-WS(4, 0.75), C=109 79% 77% * 100 cosine lr 5e-5
RandWire-WS(4, 0.75), C=78 74.7% 73.97% * 250 cosine lr 5e-5

*This result does not take advantage of dropout, droppath and label smoothing techniques. I will use these tricks to retrain the model.


  • python packages
    • pytorch = 0.4.1
    • torchvision>=0.2.1
    • tensorboardX
    • pyyaml
    • CVdevKit
    • networkx

Data Preparation

Download the ImageNet dataset and put them into the {repo_root}/data/imagenet.

Training a model from scratch

./ configs/config_regular_c109_n32.yaml


All materials in this repository are released under the Apache License 2.0.

You can’t perform that action at this time.