This directory provides information about the baseline data provided by Dopamine. We currently only support SAC for mujoco. Also, the baseline data is reported using the training regime, not evaluation. For SAC, that means we are using sampled actions, not the mean action.
The default configuration file (set up with gin configuration framework) is:
We provide a website where you can quickly visualize the training run for SAC.
The plots are rendered from a set of JSON files which we compiled. These may prove useful in their own right to compare against results obtained from other frameworks.