diff --git a/test/test_libs.py b/test/test_libs.py
index 53fc10c41fb..e6891148c2b 100644
--- a/test/test_libs.py
+++ b/test/test_libs.py
@@ -636,7 +636,7 @@ def test_vmas_spec_rollout(
         )
         for e in [env, wrapped]:
             e.set_seed(0)
-            check_env_specs(e, check_dtype=False)
+            check_env_specs(e)
             del e
 
     @pytest.mark.parametrize("num_envs", [1, 20])
diff --git a/torchrl/envs/libs/vmas.py b/torchrl/envs/libs/vmas.py
index e3880504660..0ec49794756 100644
--- a/torchrl/envs/libs/vmas.py
+++ b/torchrl/envs/libs/vmas.py
@@ -2,6 +2,7 @@
 
 import torch
 from tensordict.tensordict import TensorDict, TensorDictBase
+
 from torchrl.data import CompositeSpec, UnboundedContinuousTensorSpec
 from torchrl.envs.common import _EnvWrapper
 from torchrl.envs.libs.gym import _gym_to_torchrl_spec_transform
@@ -178,6 +179,7 @@ def _make_specs(
                             value, batch_size=torch.Size((self.num_envs,))
                         ).shape[1:],
                         device=self.device,
+                        dtype=torch.float32,
                     )
                     for key, value in self.scenario.info(agent0).items()
                 },
@@ -285,7 +287,7 @@ def read_info(self, infos: Dict[str, torch.Tensor]) -> torch.Tensor:
         infos = TensorDict(
             source={
                 key: _selective_unsqueeze(
-                    value, batch_size=torch.Size((self.num_envs,))
+                    value.to(torch.float32), batch_size=torch.Size((self.num_envs,))
                 )
                 for key, value in infos.items()
             },