keiohta / tf2rl

TensorFlow2 Reinforcement Learning
MIT License
461 stars 104 forks source link

How to use multiple GPU to train #147

Closed hongjianjiang closed 2 years ago

keiohta commented 2 years ago

I have no experience using multiple GPUs to train deep RL policies (typically the input space of RL problems is much smaller than CV setting), so I don't know how to. But I think you can do that by following some articles, such as this.

keiohta commented 2 years ago

I close this issue since no reaction from @swordJJ for several months. I believe using multiple GPUs is not directly related to tf2rl but rather related to tensorflow. If you think we need to modify codes on tf2rl side, then please open this issue. Thanks.