Train a RBM model using SINGA Tensor APIs

This example is to train a RBM model over the MNIST dataset using SINGA Python Tensor APIs. The RBM model and its hyper-parameters are set following Hinton's paper

Running instructions

1. Download the pre-processed MNIST dataset

2. Start the training

    python train.py mnist.pkl.gz
   

By default the training code would run on CPU. To run it on a GPU card, please start the program with an additional argument

    python train.py mnist.pkl.gz --use_gpu