MTSAC implementation based on Metaworld. Benchmarked on MT1.

This experiment shows how MTSAC adapts to 50 environents of the same type but each environment has a goal variation.

mtsac_metaworld_mt1_pick_place(ctxt=None, *, seed, timesteps, _gpu)

Train MTSAC with the MT1 pick-place-v1 environment.

  • 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.

  • _gpu (int) – The ID of the gpu to be used (used on multi-gpu machines).

  • timesteps (int) – Number of timesteps to run.