This repositary contains a simple example to get in touch with Tensorflow and a toy example which trains with different models a classificator with MNIST dataset.
In the file simple_network.py
you will find an easy example on how to make a neural network with three fully-connected layers. Using a random input of 0 and 1, the script trains the model and provides with the accuracy of the system.
In the file train_mnist.py
you will find an example of how to train and classify the MNIST dataset. Within the code you will find two different architectures. The first one which consist on three fully-connected layers and the second one which consist on the LeNet architecture. To use the latter set the variable of the main script:
lenet_model = 1
This sript allows you to visualize in real time the training with Tensorboard.