The A3C algorithm implemented using Pytorch and heavily inspired by the implementation found at ikostrikov's repo. Built for a Reinforcement Learning course at Oxford University.
- Implement command line arguments
- Implement resume training feature
- Make training processes exit upon convergence
- Make testing process exit when training process exits
- Implement logging (tensorboard/visdom)