Add rollout_buffer_class and rollout_buffer_kwargs to PPO.

stable_baselines3/ppo/ppo.py

@@ -52,6 +52,8 @@ class PPO(OnPolicyAlgorithm):
         instead of action noise exploration (default: False)
     :param sde_sample_freq: Sample a new noise matrix every n steps when using gSDE
         Default: -1 (only sample at the beginning of the rollout)
+    :param rollout_buffer_class: Rollout buffer class to use. If ``None``, it will be automatically selected.
+    :param rollout_buffer_kwargs: Keyword arguments to pass to the rollout buffer on creation
     :param target_kl: Limit the KL divergence between updates,
         because the clipping is not enough to prevent large update
         see issue #213 (cf https://github.com/hill-a/stable-baselines/issues/213)
@@ -92,6 +94,8 @@ def __init__(
         max_grad_norm: float = 0.5,
         use_sde: bool = False,
         sde_sample_freq: int = -1,
+        rollout_buffer_class: Optional[Type[RolloutBuffer]] = None,
+        rollout_buffer_kwargs: Optional[Dict[str, Any]] = None,
         target_kl: Optional[float] = None,
         stats_window_size: int = 100,
         tensorboard_log: Optional[str] = None,
@@ -113,6 +117,8 @@ def __init__(
             max_grad_norm=max_grad_norm,
             use_sde=use_sde,
             sde_sample_freq=sde_sample_freq,
+            rollout_buffer_class=rollout_buffer_class,
+            rollout_buffer_kwargs=rollout_buffer_kwargs,
             stats_window_size=stats_window_size,
             tensorboard_log=tensorboard_log,
             policy_kwargs=policy_kwargs,