| 21 | |
| 22 | |
| 23 | class PpoBufferSamples(NamedTuple): |
| 24 | observations: Dict[str, th.Tensor] |
| 25 | actions: th.Tensor |
| 26 | old_values: th.Tensor |
| 27 | old_log_prob: th.Tensor |
| 28 | old_mu: th.Tensor |
| 29 | old_sigma: th.Tensor |
| 30 | advantages: th.Tensor |
| 31 | returns: th.Tensor |
| 32 | exploration_suggests: List[tuple] |
| 33 | |
| 34 | |
| 35 | class PpoBuffer(): |