chrisabbott / alexnet Goto Github PK

Section (0): REQUIREMENTS FOR RUNNING OUR PROGRAM

To run our program, you will need to have Python 3.6 and the
ubiquitous "virtualenv" installed on your system. Besides
that, our Makefile will handle all other requirements.

We have provided an easy-to-use Makefile to help you run our
program.

------------------------------------------------------------

Section (1): TRAINING A TINY IMAGENET CLASSIFIER FROM SCRATCH

To train a new classifier, "cd" into our submission
directory and execute the following commands:

make prepare-venv
source env/bin/activate
python train.py &
python evaluate.py &
tensorboard --logdir=logs

Then, open TensorBoard in your browser to monitor the
classifier's accuracy as it is being trained. You can
also monitor loss, mean-squared error, and other metrics.

------------------------------------------------------------

Section (2): CONTINUE TRAINING OUR ALREADY-TRAINED CLASSIFIER

As part of our submission, we included a way to test the
accuracy of our already-trained classifier. Execute the
following commands in a new terminal:

make prepare-venv
source env/bin/activate
python train.py $(pwd)/trained-logs/ &
python evaluate.py $(pwd)/trained-logs/ &
tensorboard --logdir=trained-logs

^^^ TODO TODO TODO TODO TODO ^^^

Like (1), you can open TensorBoard to monitor the
classifier's accuracy.