Update to Gymnasium 1.2.0 (#2150)

* Update to Gymnasium 1.2.0

* Fix for new gym version

Author: araffin

docs/guide/examples.rst

@@ -746,6 +746,7 @@ you can find links to some of these wrappers in `issue #772 <https://github.com/
 - Isaac Lab wrapper: `link <https://github.com/isaac-sim/IsaacLab/blob/main/source/extensions/omni.isaac.lab_tasks/omni/isaac/lab_tasks/utils/wrappers/sb3.py>`__
 - Brax: `link <https://gist.github.com/araffin/a7a576ec1453e74d9bb93120918ef7e7>`__
 - EnvPool: `link <https://github.com/sail-sg/envpool/blob/main/examples/sb3_examples/ppo.py>`__
+- Getting SAC to Work on a Massive Parallel Simulator: https://araffin.github.io/post/sac-massive-sim/
 
 
 SB3 with DeepMind Control (dm_control)

docs/misc/changelog.rst

@@ -3,7 +3,7 @@
 Changelog
 ==========
 
-Release 2.7.0a0 (WIP)
+Release 2.7.0a1 (WIP)
 --------------------------
 
 Breaking Changes:
@@ -13,6 +13,7 @@ New Features:
 ^^^^^^^^^^^^^
 - Added support for n-step returns for off-policy algorithms via the `n_steps` parameter
 - Added ``NStepReplayBuffer`` that allows to compute n-step returns without additional memory requirement (and without for loops)
+- Added Gymnasium v1.2 support
 
 Bug Fixes:
 ^^^^^^^^^^

setup.py

@@ -76,7 +76,7 @@
     packages=[package for package in find_packages() if package.startswith("stable_baselines3")],
     package_data={"stable_baselines3": ["py.typed", "version.txt"]},
     install_requires=[
-        "gymnasium>=0.29.1,<1.2.0",
+        "gymnasium>=0.29.1,<1.3.0",
         "numpy>=1.20,<3.0",
         "torch>=2.3,<3.0",
         # For saving models

stable_baselines3/common/results_plotter.py

@@ -24,7 +24,7 @@ def rolling_window(array: np.ndarray, window: int) -> np.ndarray:
     :param window: length of the rolling window
     :return: rolling window on the input array
     """
-    shape = array.shape[:-1] + (array.shape[-1] - window + 1, window)
+    shape = array.shape[:-1] + (array.shape[-1] - window + 1, window)  # noqa: RUF005
     strides = (*array.strides, array.strides[-1])
     return np.lib.stride_tricks.as_strided(array, shape=shape, strides=strides)
 

stable_baselines3/version.txt

@@ -1 +1 @@
-2.7.0a0
+2.7.0a1

tests/test_vec_normalize.py

@@ -180,7 +180,11 @@ def _make_warmstart(env_fn, **kwargs):
 
 def _make_warmstart_cliffwalking(**kwargs):
     """Warm-start VecNormalize by stepping through CliffWalking"""
-    return _make_warmstart(lambda: gym.make("CliffWalking-v0"), **kwargs)
+    try:
+        return _make_warmstart(lambda: gym.make("CliffWalking-v0"), **kwargs)
+    except gym.error.DeprecatedEnv:
+        # v1 required since Gymnasium v1.2.0
+        return _make_warmstart(lambda: gym.make("CliffWalking-v1"), **kwargs)
 
 
 def _make_warmstart_cartpole():