Step 1: create an vir env in the requirement.txt 
Step 2: install baseline 
Step 3: run the following command.


####### Learning curves to evaluate HNS: three seeds and five algorithms in each game across 55 games

# python main.py --game breakout --method DQN --seed 1 --gpu 1
# python main.py --game breakout --method C51 --seed 1 --gpu 1
# python main.py --game breakout --method QRDQN --seed 1 --gpu 1
# python main.py --game breakout --method MMD --seed 1 --gpu 1
# python main.py --game breakout --method Sinkhorn --niter_sink 10 --epsilon 10 --seed 1 --gpu 1

####### Multi-dimensioanl reward setting: use the reward-compose file, two algorithm: Sinkhorn, MMD

python main.py --game Asteroids --method Sinkhorn --iter 10000000 --multi 1 --seed 1 --gpu 1 
python main.py --game AirRaid --method Sinkhorn --iter 10000000 --multi 1 --seed 1 --gpu 1 
python main.py --game MsPacman --method Sinkhorn --iter 10000000 --multi 1 --seed 1 --gpu 1 
python main.py --game UpNDown --method Sinkhorn --iter 10000000 --multi 1 --seed 1 --gpu 1 
python main.py --game Gopher --method Sinkhorn --iter 10000000 --multi 1 --seed 1 --gpu 1 
python main.py --game pong --method Sinkhorn --iter 10000000 --multi 1 --seed 1 --gpu 1