(self)
| 75 | return sum_independent_dims(log_prob) |
| 76 | |
| 77 | def entropy_loss(self) -> th.Tensor: |
| 78 | entropy_loss = -1.0 * self.distribution.entropy() |
| 79 | return th.mean(entropy_loss) |
| 80 | |
| 81 | def exploration_loss(self, exploration_suggests) -> th.Tensor: |
| 82 | # [('stop'/'go'/None, 'turn'/'straight'/None)] |