trpo_gym_tf_cartpole_pretrained

An example to train a task with TRPO algorithm.

trpo_gym_tf_cartpole(ctxt=None, seed=1)

Train TRPO with CartPole-v0 environment.

Parameters

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

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

pre_trained_trpo_cartpole(ctxt=None, snapshot_dir='data/local/experiment/trpo_gym_tf_cartpole', seed=1)

Use pre-trained TRPO and reusume experiment.

Parameters

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

• snapshot_dir (path) – directory to snapshot

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

log_dir = data/local/experiment/trpo_gym_tf_cartpole