DLBeginnerExercises

Deep Learning Exercises for Beginners using Keras (Tensorflow)

Setup

• Anaconda (conda 4.7.11)
• Cuda (V10.0.130)
• Python 3.6

Create new environment

conda create -n keras-gpu python=3.6 numpy scipy keras-gpu

Activate new env

conda activate keras-gpu

Test run setup

Model: "sequential"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
dense (Dense)                (None, 512)               401920    
_________________________________________________________________
dropout (Dropout)            (None, 512)               0         
_________________________________________________________________
dense_1 (Dense)              (None, 512)               262656    
_________________________________________________________________
dropout_1 (Dropout)          (None, 512)               0         
_________________________________________________________________
dense_2 (Dense)              (None, 10)                5130      
=================================================================
Total params: 669,706
Trainable params: 669,706
Non-trainable params: 0
_________________________________________________________________
Train on 60000 samples, validate on 10000 samples

Run mnist_mlp.py

python mnist_mlp.py

Sample Output

mnist_mlp_run.txt

After verifying your setup, try this tutorial: