carpedm20 / deep-rl-tensorflow

TensorFlow implementation of Deep Reinforcement Learning papers
MIT License
1.6k stars 396 forks source link

returns cumulated reward for each step #13

Closed dillonalaird closed 7 years ago

dillonalaird commented 7 years ago

Step should return the total reward received over n_action_repeat instead of the last reward received.

carpedm20 commented 7 years ago

I think it's better to make this changeable depends on the environment.