To train the agent:

python3 main_keras.py --dueling true --train_dqn

To run the code with a trained agent:

python3 main_keras.py --dueling true --test_dqn_model_path saved_dqn_networks/DDDQN_Litter_Agent_2420000.h5 --test_dqn

for visual explanations you need scipy=1.2.0

python3 main_keras.py --dueling true --test_dqn_model_path saved_dqn_networks/DDDQN_Litter_Agent_2420000.h5 --test_dqn --visualize --num_frames 300

Name		Name	Last commit message	Last commit date
Latest commit History 25 Commits
agents		agents
dqn_summary		dqn_summary
visualization		visualization
.gitignore		.gitignore
README.md		README.md
argument.py		argument.py
distributions.py		distributions.py
envs.py		envs.py
keras_model.py		keras_model.py
main.py		main.py
main_keras.py		main_keras.py
model.py		model.py
model_plot.png		model_plot.png
myopic.py		myopic.py
print_runs.py		print_runs.py
visprocess.py		visprocess.py