Skip to content
himanshusahni worker name scope should have trailing backslash otherwise any worker…
… 10-19 will clash in scope with worker 1, and so on.
Latest commit bc7ee05 Oct 10, 2017

README.md

Implementation of A3C (Asynchronous Advantage Actor-Critic)

Running

./train.py --model_dir /tmp/a3c --env Breakout-v0 --t_max 5 --eval_every 300 --parallelism 8

See ./train.py --help for a full list of options. Then, monitor training progress in Tensorboard:

tensorboard --logdir=/tmp/a3c

Components

  • train.py contains the main method to start training.
  • estimators.py contains the Tensorflow graph definitions for the Policy and Value networks.
  • worker.py contains code that runs in each worker threads.
  • policy_monitor.py contains code that evaluates the policy network by running an episode and saving rewards to Tensorboard.
You can’t perform that action at this time.