First off want to say big fan of pytorch and thanks to all the pytorch guys for making such a great library for neural networks. And the above and beyond attentiveness to all the help I see you provide in this forum. Awesome!
Just wanted to share my recent reinforcement learning repo with the A3C algorithm and training agent in Atari in openai gym. Only trained agent in Pong-v0, MsPacman-v0, and Breakout-v0 so far and had pretty good success so far with agent getting top scores on each in openai gym environment leaderboards. Probably will train in a couple more games but another shining example of pytorch’s stellar performance.
Any feedback or questions always welcome