Pytorch-XNOR-Net

Build

cd <Repository Root>/csrc/binop
make

MNIST

Usage

Train:

cd <Repository Root>/MNIST/
python3 main.py --arch Bin_LeNet
python3 main.py --arch LeNet

Evaluate:

cd <Repository Root>/MNIST/
python3 main.py --arch Bin_LeNet --evaluate --pretrained ./models/Bin_LeNet.best.pth # --no_cuda (Use CPU)python3 main.py --arch LeNet --evaluate --pretrained ./models/LeNet.best.pth # --no_cuda (Use CPU)

Result

Cifar10

Usage

Train:

cd <Repository Root>/Cifar10/
python3 main.py --arch Bin_VGG16 #(11, 13, 16, 19)python3 main.py --arch VGG16 #(11, 13, 16, 19)

Evaluate:

cd <Repository Root>/Cifar10/
python3 main.py --arch Bin_VGG16 --evaluate --pretrained ./models/Bin_VGG16.best.pth # --no_cuda (Use CPU)python3 main.py --arch VGG16 --evaluate --pretrained ./models/VGG16.best.pth # --no_cuda (Use CPU)

Result

Pre-trained models

Google Drive

Environment

Software

• Ubuntu 16.04

• Python 3.5

• Pytorch 0.3.1

• CUDA 8.0

• gcc 5.4

Hardware

• NVIDIA GTX 1080

• Intel i5-6500 CPU @ 3.20GHz × 4

Reference

• Binarized Neural Networks: Training Deep Neural Networks with Weights and Activations Constrained to +1 or -1

• XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks

• https://github.com/jiecaoyu/XNOR-Net-PyTorch

• cpu-gemm

• cpu-conv2d

• gpu-gemm and gpu-conv2d

• popcount