BNN-XNOR-BiReal

This is a project contains Pytorch implementations of the BNN*, XNOR-Net and BiReal-Net. (NOTE: the first and last layer of BNN* is floating point.)

---

Results

Method	Acc
BNN	71.70%
XNOR	80.77%
BiReal	83.17%

Dependencies

• Python 3.5
• Pytorch 1.3
• Cuda

Train

• Train a binary model with bireal:

python train.py --binary_method bireal

• Train a binary model with xnor:

python train.py --binary_method xnor

• Train a binary model with bnn:

python train.py --binary_method bnn

Todo

• Pretrain model on cifar10

Reference

• XNOR-Net-Pytorch
• Bi-Real-net
• BinaryNet: Training Deep Neural Networks with Weights and Activations Constrained to +1 or −1
• XNOR-Net: ImageNet Classification Using Binary Convolutional Neural Networks
• Bi-Real Net: Enhancing the Performance of 1-bit CNNs With Improved Representational Capability and Advanced Training Algorithm