jmduarte/mnist-training

MNIST MLP

Quick Start

To setup with conda/pip (assuming they already installed):

source setup.sh

To train the model and generate .h5 and .json files:

make train

To test the model:

make prediction

To prune the model to 90% sparsity:

make prune

To check the model compression:

python check_compression.py

To run a hyperparameter scan with guild.ai:

guild run train epochs=100 optimizer=[adam,nadam,rmsprop,sgd] dropout_rate=[0,0.1,0.2] l1_reg=[0,1e-5,1e-4]

To run different pruning hyperparameters:

guild run prune epochs=100 optimizer=[adam,nadam,rmsprop,sgd] final_sparsity=[0.3,0.5,0.7,0.9,0.95,0.97,0.99]