EECS6322_PROJECT

Steps To Run

Step 1. Create a new python environment in the current directory using the command:

python -m venv .

Step 2. Activate the environment using the following commands on windows:

cd scripts
activate

and on linux

source bin/activate

Step 3. Install the requirements using the following command:

pip install -r requirements.txt

and install pytorch directly from source

pip install torch==1.6.0+cu101 torchvision==0.7.0+cu101 -f https://download.pytorch.org/whl/torch_stable.html

Step 4: Add model checkpoint directory

mkdir model_checkpoints

and point the config.ini file to this folder:

checkpoint_dir = model_checkpoints/

Step 5. Train stage 1

First stage of model should be run for 10 epochs, batch size 6. Check the config.ini file to verify this:

epochs = 10
batch_size = 6
stage = 1

Then run the train.py python file to start training using the following command:

python train.py --config_file config.ini --section DEFAULT

Step 6. Train stage 2

Second stage of model should be run for 15 epochs, batch size 3. Change the config.ini file to reflect this:

epochs = 15
batch_size = 3
stage = 2
model_stage1_file = C2F_stage1_epoch9.pt

Then run the train.py python file to start training again using the following command:

python train.py --config_file config.ini --section DEFAULT

Step 7. Test Model

To test the saved model, point the config.ini file to the final saved model:

model_test_file = C2F_stage2_epoch14.pt

Then run the test.py python file to run test phase using the following command:

python test.py --config_file config.ini --section DEFAULT

Generated summaries will be the GPT_C2F_output.json file in the output folder specified by config.ini file:

test_output_dir = output/

Steps To Run Stretch Project

Repeat steps 5-7 but replace train.py and test.py with train_stretch.py and test_stretch.py