Make play_random to fill memory

tambetm / simple_dqn

Simple deep Q-learning agent.

MIT License

692 stars 184 forks source link

Closed mthrok closed 8 years ago

mthrok commented 8 years ago

play_random function, which is called before training to populate replay memory was not actually adding replay to ReplayMemory. This PR fixes it.

play_random function is not used anywhere else, so this change should not have any side effect. https://github.com/tambetm/simple_dqn/search?utf8=✓&q=play_random

tambetm commented 8 years ago

Thanks @mthrok, I overlooked that when I was adding the StateBuffer.