Awesome

MNIST with TensorFlow

• Just for training TensorFlow and Deep Learning
• Try to make easy to understand building layers and using TensorFlow
  • write summaries for TensorBoard
  • save and load a model and reuse for prediction
• Pre-trained model with default options is included
  • you can test prediction and TensorBoard without any hassle

Class

• MNIST : building model (currently CNN only)
• MNISTTrainer : training logic and steps
• MNISTTester : test trained model and an image
• TFUtils : Xavier initialization and a small utilities for my laziness

Excutable Scripts

• train.py : can use below options
  • learning_rate=0.001
  • decay=0.9
  • training_epochs=10
  • batch_size=100
  • p_keep_conv=0.8
  • p_keep_hidden=0.5
• test.py
  • prediction test with MNIST test set
  • prediction test with image file
    • only for square images and single number
    • size is not matter

Results

➜  TensorFlow-MNIST# python train.py 
Preparing MNIST data..
Extracting mnist/data/train-images-idx3-ubyte.gz
Extracting mnist/data/train-labels-idx1-ubyte.gz
Extracting mnist/data/t10k-images-idx3-ubyte.gz
Extracting mnist/data/t10k-labels-idx1-ubyte.gz
---
Building CNN model..
---
Start training. Please be patient. :-)
Epoch: 0001 / Accuracy = 0.9511
Epoch: 0002 / Accuracy = 0.9634
...
---
Saving my model..
---
Learning Finished!

➜  TensorFlow-MNIST# python test.py
---
Loading a model..
Preparing MNIST data..
Extracting mnist/data/train-images-idx3-ubyte.gz
Extracting mnist/data/train-labels-idx1-ubyte.gz
Extracting mnist/data/t10k-images-idx3-ubyte.gz
Extracting mnist/data/t10k-labels-idx1-ubyte.gz
---
Calculating accuracy of test set..
---
CNN accuracy of test set: 0.993600
---
Predict random item: 5 is 5, accuracy: 1.000
---
4 is digit-4.png, accuracy: 1.000000
---
2 is digit-2.png, accuracy: 1.000000
---
5 is digit-5.png, accuracy: 0.997631

➜  TensorFlow-MNIST# tensorboard --logdir=logs/mnist-cnn