Skip to content

Latest commit

 

History

History
36 lines (25 loc) · 855 Bytes

README.md

File metadata and controls

36 lines (25 loc) · 855 Bytes

cnn-mnist

This is a Python3 / Tensorflow implementation of a convolutional network model for MNIST dataset.

Setup

To run this code you need the following:

  • Python3.3+
  • Numpy
  • TensorFlow

Training the model

Use the main.py script to train the model. To train the default model on MNIST simply use:

python3 main.py

This code will automatically output the predictions of test-image, which is in the mnist_data/ folder.

To use the pretrained model append the -l flag after the command.

To only predict the test images using the pretrained model append the -l -e 0 flag after the command.

Performance

Train/Validation: 60000/5000

Train the model for 30 epochs.

Data Accuarrcy
train 0.9989
validation 0.9932