{"full_name":"DLR-RM/stable-baselines3","owner":"DLR-RM","name":"stable-baselines3","shard":"1","url":"https://github.com/DLR-RM/stable-baselines3","description":"PyTorch version of Stable Baselines, reliable implementations of reinforcement learning algorithms. ","homepage":"https://stable-baselines3.readthedocs.io","topics":["reinforcement-learning","reinforcement-learning-algorithms","machine-learning","gym","openai","baselines","toolbox","stable-baselines","python","pytorch","robotics","sde","gsde","sb3"],"tag_pages":["gym","machine-learning","openai","python","pytorch","reinforcement-learning","robotics","toolbox"],"language":"Python","license":"MIT","avatar":"https://avatars.githubusercontent.com/u/25227811?v=4","created_at":"2020-05-05T05:52:26Z","pushed_at":"2026-10-26T09:35:00Z","metrics":{"stars":14015,"forks":2183,"watchers":68,"issues":1605,"prs":666,"releases":33,"issues_open":59,"issues_closed":1546,"prs_open":30,"prs_closed":124,"prs_merged":512,"discussions":0},"growth":{"day":{"stars":3,"forks":0,"base_day":"2026-11-15"},"week":{"stars":57,"forks":3,"base_day":"2026-11-04"},"month":{"stars":114,"forks":6,"base_day":"2026-10-17"}},"history":{"dates":["2026-09-06","2026-09-07","2026-09-08","2026-09-09","2026-09-10","2026-09-11","2026-09-12","2026-09-13","2026-09-14","2026-09-15","2026-09-16","2026-09-17","2026-09-18","2026-09-19","2026-09-20","2026-09-21","2026-09-22","2026-09-23","2026-09-24","2026-09-25","2026-09-26","2026-09-27","2026-09-28","2026-09-29","2026-09-30","2026-10-01","2026-10-02","2026-10-03","2026-10-04","2026-10-05","2026-10-06","2026-10-07","2026-10-08","2026-10-09","2026-10-10","2026-10-11","2026-10-12","2026-10-13","2026-10-14","2026-10-15","2026-10-16","2026-10-17","2026-10-18","2026-10-19","2026-10-20","2026-10-21","2026-10-22","2026-10-23","2026-10-24","2026-10-25","2026-10-26","2026-10-27","2026-10-28","2026-10-29","2026-10-30","2026-10-31","2026-11-01","2026-11-02","2026-11-03","2026-11-04","2026-11-10","2026-11-11","2026-11-12","2026-11-13","2026-11-14","2026-11-15","2026-11-16"],"stars":[13770,13774,13776,13776,13779,13780,13781,13786,13790,13796,13802,13810,13818,13819,13821,13823,13829,13837,13843,13849,13850,13850,13849,13855,13861,13862,13862,13860,13863,13864,13868,13871,13875,13878,13880,13881,13883,13885,13889,13897,13898,13901,13902,13904,13910,13914,13916,13922,13925,13929,13934,13937,13939,13942,13946,13952,13953,13956,13957,13958,13999,14004,14008,14008,14009,14012,14015],"forks":[2177,2176,2177,2177,2178,2179,2179,2179,2180,2179,2180,2181,2182,2182,2182,2182,2182,2182,2182,2183,2184,2184,2184,2183,2183,2182,2183,2183,2183,2183,2183,2183,2183,2181,2179,2178,2178,2177,2176,2176,2178,2177,2178,2177,2177,2177,2177,2177,2179,2179,2179,2180,2178,2177,2178,2179,2180,2179,2179,2180,2183,2183,2183,2183,2183,2183,2183],"watchers":[67,68,68,68,68,68,68,68,68,68,69,69,69,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,70,68,68,68,68,68,68,68,68,68,68,68,null,68,68,68,68,68,68],"issues_open":[null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,59,59,59,59,59,59],"prs_open":[null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,null,30,30,30,30,30,30]},"boards":{},"same_owner":[],"issues":[{"number":635,"title":"AtariWrapper does not use recommended defaults","comments":24,"created_at":"2021-10-27T21:59:45Z"},{"number":1852,"title":"`torch.load` without `weights_only` parameter is unsafe","comments":18,"created_at":"2024-02-27T20:44:09Z"},{"number":914,"title":"Supporting PyTorch GPU compatibility on Apple Silicon chips","comments":17,"created_at":"2022-05-20T18:33:49Z"},{"number":1692,"title":"[Bug]: Rendering with EvalCallback does not render the initial or final state","comments":14,"created_at":"2023-09-23T23:49:53Z"},{"number":1630,"title":"[Question] I do not understand the GPU and memory usage of SB3","comments":13,"created_at":"2023-07-27T08:56:39Z"},{"number":338,"title":"[Feature Request] independently configurable learning rates for actor and critic","comments":11,"created_at":"2021-03-03T15:20:24Z"},{"number":1345,"title":"[Feature Request] Multi-Cluster Support for SubprocVecEnv","comments":10,"created_at":"2023-02-23T14:17:13Z"},{"number":1242,"title":"Prioritized Experience Replay for DQN","comments":10,"created_at":"2022-12-26T10:24:19Z"}],"discussions":[],"latest_release":{"name":"v2.9.2: Bug fixes and doc improvements","tag":"v2.9.2","published_at":"2026-10-26T09:35:00Z","body":"> [!WARNING]\r\n> Stable-Baselines3 (SB3) v2.9.2 will be the last one supporting Python 3.10 (end of life in October 2026)\r\n> We highly recommended you to upgrade to Python >= 3.11.\r\n\r\nSB3 Contrib (more algorithms): https://github.com/Stable-Baselines-Team/stable-baselines3-contrib\r\nRL Zoo3 (training framework): https://github.com/DLR-RM/rl-baselines3-zoo\r\nStable-Baselines Jax (SBX): https://github.com/araffin/sbx\r\n\r\nTo upgrade:\r\n```\r\npip install stable_baselines3 sb3_contrib rl_zoo3 --upgrade\r\n```\r\n\r\n\r\n### Bug Fixes:\r\n\r\n- Fixed file descriptor leak in `SubprocVecEnv.close()` by closing parent-side pipes to prevent \"too many open files\" errors in long-running loops\r\n- Fixed Pandas deprecation warnings on `inplace` operations\r\n\r\n### [SB3-Contrib](https://github.com/Stable-Baselines-Team/stable-baselines3-contrib)\r\n- Upgraded to Stable-Baselines3 >= 2.9.2\r\n\r\n### [RL Zoo](https://github.com/DLR-RM/rl-baselines3-zoo)\r\n- Set `learning_starts` to 500 for `parking-v0` in TQC hyperparameters to ensure the first episode is finished\r\n- Fixed \"too many open files\" issue when doing hyperparameter optimization: the trial environments are now closed in a `finally` block and `gc_after_trial=True` is passed to Optuna to free them between trials (also needs latest SB3 master version to fully work)\r\n- Set `use_expln=True` for pybullet envs when using gSDE to avoid NaNs\r\n- Add a note regarding pybullet source installation error on macOS due to legacy code\r\n- Migrate packaging setup from `setup.py` to `pyproject.toml` and integrate `requirements.txt` as a new extra\r\n\r\n### [SBX (SB3 + Jax)](https://github.com/araffin/sbx)\r\n- Updated Jax dependency version range\r\n\r\n### Deprecations:\r\n\r\n### Others:\r\n\r\n- Fixed moviepy test compatibility with moviepy >= 2.0 (PyTorch doesn't support it yet)\r\n- Added filter for numpy `newshape` deprecation warning from torch tensorboard\r\n- Fixed mypy error in test_vec_envs.py by wrapping `itertools.product` with `list()`\r\n- Improved tests coverage for RMSpropTFLike optimizer (invalid params, centered/momentum/weight_decay branches, pickle roundtrip)\r\n- Improved test coverage for `save_util.py`: added tests for `BadZipFile` error handling in `load_from_zip_file` and `IsADirectoryError`/`FileNotFoundError` handling in `open_path`\r\n- Fixed Docker build by adding the missing `--system` flag to `uv pip uninstall opencv-python` (required by recent `uv`)\r\n- Added a warning when `n_steps > 1` is passed together with a custom `replay_buffer_class`, as `n_steps` is only applied when the replay buffer class is selected automatically (@Koustav-github)\r\n- Updated tests dependencies\r\n\r\n### Documentation:\r\n\r\n- Clarified in the `n_steps` docstring that it only applies when `replay_buffer_class` is `None`, and that a custom replay buffer must be configured with `replay_buffer_kwargs={\"n_steps\": ..., \"gamma\": ...}` (@Koustav-github)\r\n- Updated algorithm table (added the ones from SBX and sorted the table)\r\n- Fixed typos, grammar, and Gymnasium API inconsistencies across documentation\r\n\r\n## New Contributors\r\n* @nikolauspschuetz made their first contribution in https://github.com/DLR-RM/stable-baselines3/pull/2272\r\n* @Koustav-github made their first contribution in https://github.com/DLR-RM/stable-baselines3/pull/2275\r\n\r\n**Full Changelog**: https://github.com/DLR-RM/stable-baselines3/compare/v2.9.0...v2.9.2","is_prerelease":false},"readme":true,"readme_zh":false,"news":[],"stale":false,"snapshot_at":"2026-11-16T09:23:14Z","history_start":"2026-09-06"}