pipeline.txt

The model takes input from training and test data and outputs the result 
in the file named 'submission_simple_keras_starter.csv'.

We ran out model in Kaggle. We used keras to implement our model.

Our model has an inception layer at first. This layer merges three convolutional 
layers who has 64 filters each and whose filter sizes are 1*1, 3*3, and 5*5.

Then we have 8 convolutional layers with maxpool layers. 

First two convolutional layers have 32 filters of size 5*5.
Then we added a max-pool layer.

The next two convolutional layers have 64 filters of size 3*3.
Then we added a max-pool layer.

The next two convolutional layers have 128 filters of size 3*3.
Then we added a max-pool layer.

The next two convolutional layers have 256 filters of size 3*3.
Then we added a max-pool layer.

After that we added two dense layers of size 64 and 10.