Fixed DictRolloutBuffer dtype assignment

Trenza1ore · Trenza1ore · commit 66f830086b1b · 2025-07-27T13:35:49.000+01:00
diff --git a/stable_baselines3/common/buffers.py b/stable_baselines3/common/buffers.py
@@ -788,7 +788,7 @@ def reset(self) -> None:
         self.observations = {}
         for key, obs_input_shape in self.obs_shape.items():
             self.observations[key] = np.zeros(
-                (self.buffer_size, self.n_envs, *obs_input_shape), dtype=self.dtypes["observations"]
+                (self.buffer_size, self.n_envs, *obs_input_shape), dtype=self.dtypes["observations"][key]
             )
         self.actions = np.zeros((self.buffer_size, self.n_envs, self.action_dim), dtype=self.dtypes["actions"])
         self.rewards = np.zeros((self.buffer_size, self.n_envs), dtype=np.float32)

Original file line number	Diff line number	Diff line change
`@@ -788,7 +788,7 @@ def reset(self) -> None:`
`788`	`788`	`self.observations = {}`
`789`	`789`	`for key, obs_input_shape in self.obs_shape.items():`
`790`	`790`	`self.observations[key] = np.zeros(`
`791`		`- (self.buffer_size, self.n_envs, *obs_input_shape), dtype=self.dtypes["observations"]`
	`791`	`+ (self.buffer_size, self.n_envs, *obs_input_shape), dtype=self.dtypes["observations"][key]`
`792`	`792`	`)`
`793`	`793`	`self.actions = np.zeros((self.buffer_size, self.n_envs, self.action_dim), dtype=self.dtypes["actions"])`
`794`	`794`	`self.rewards = np.zeros((self.buffer_size, self.n_envs), dtype=np.float32)`