This codebase has been tested with python 3.5 and 3.6.
To install: clone the repository and run
pip install -r requirements.py
Add the repo to you pythonpath by running `export PYTHONPATH=[path/to/repo/]cpv/crafting:$PYTHONPATH'
To generate training data for the crafting environment, run
cd [path/to/repo/]cpv/
mkdir data
python crafting/scripts/collect_composite_trajectories.py
This may take a while.
Then, collect the evaluation trajectories by running
python crafting/scripts/collect_reference_trajectories.py
To train a CPV-Full model, run
mkdir octresults
python crafting/gridworld/algorithms/cpv_experiment.py -H -P
This script will print where the checkpoints are saved and where the tensorboard logs are saved.
To evaluate the model online, run
python crafting/scripts/run_model_multitask_tensorboard.py --model [path/to/checkpoints_dir] --tb [path/to/tensorboard_dir] --type V3