maml_trpo_half_cheetah_dir

This is an example to train MAML-VPG on HalfCheetahDirEnv environment.

maml_trpo_half_cheetah_dir(ctxt, seed, epochs, episodes_per_task, meta_batch_size)

Set up environment and algorithm and run the task.

Parameters
  • ctxt (ExperimentContext) – The experiment configuration used by Trainer to create the Snapshotter.

  • seed (int) – Used to seed the random number generator to produce determinism.

  • epochs (int) – Number of training epochs.

  • episodes_per_task (int) – Number of episodes per epoch per task for training.

  • meta_batch_size (int) – Number of tasks sampled per batch.