An unofficial Tensorflow implementation for the paper BinaryConnect: Training Deep Neural Networks with binary weights during propagations.

• Tensorflow 1.0
• Python 3.5
• six

The code was tested on windows 7/8.1/10.

To train a model run runme_train.py.

• Toggle binary training using --binary=True/False.
• Toggle stochastic training using --stochastic=True/False.

CIFAR-10 dataset will be automatically downloaded to ./dataset folder.

For additional configurations run python run_me.py -h.

To test a model run runme_test.py. Pass a trained model path (ckpt file) by setting --model_path.