(steps=100)
| 759 | |
| 760 | |
| 761 | def simple_rollout(steps=100): |
| 762 | # preallocate: |
| 763 | data = TensorDict({}, [steps]) |
| 764 | # reset |
| 765 | _data = env.reset() |
| 766 | for i in range(steps): |
| 767 | _data["action"] = env.action_spec.rand() |
| 768 | _data = env.step(_data) |
| 769 | data[i] = _data |
| 770 | _data = step_mdp(_data, keep_other=True) |
| 771 | return data |
| 772 | |
| 773 | |
| 774 | print("data from rollout:", simple_rollout(100)) |