search

google-deepmind / dqn_zoo

DQN Zoo is a collection of reference implementations of reinforcement learning agents developed at DeepMind based on the Deep Q-Network (DQN) agent.
Apache License 2.0
451 stars 78 forks source link

vectored environments #16

Closed raymondchua closed 2 years ago

raymondchua commented 2 years ago

Hi, Does this work with vectorized environments (such as using gym.vector.AsyncVectorEnv function)?

Best, Raymond

jqdm commented 2 years ago

Not out of the box since DQN and variants expect a single stream of experience. One could get something working reasonably quickly by adapting TransitionAccumulator so it accepts a batch of time steps.

raymondchua commented 2 years ago

Awesome, thanks @jqdm !