v6.0.0

6.0.0 (2024-03-15)

⚠ BREAKING CHANGES

• buffers: the ReplayBuffer and TrajectoryBuffer classes don't take a rew_dim argument anymore.
• package: The package should now be used as stable_learning_control instead of bayesian_learning_control.
• This package now depends on Gymnasium instead of Gym. Furthermore, it also requires Gymnasium>=26 (see https://gymnasium.farama.org/content/migration-guide/) for more information.
• The simzoo package is no longer included and has to be installed separately through the stable-gym repository.

Features

• add docker container (#379) (972079f)
• add hyperparameter variant support for exp config files (6ea47f0)
• add small policy test script (c701d69)
• add support for dictionary type observation spaces (e3bf761)
• add torch reproducibility code (#320) (89ef5a2)
• enable GPU device selection (#406) (73c1374)
• exp_cfg: fix 'start_policy' empty config bug (e7f3cf9)
• exp: add lambda lr check experiments (fedfa16)
• exp: add min statistics to rstaa2024 data analysis (16ff564)
• exp: add rstaa 2024 data analysis script (f0df0a4)
• exp: rename CompOscillator lambda experiment (ec7103c)
• improve 'eval_robustness' utility (#313) (3985867)
• improve hyperparmeter tuning, logger and add W&B logging (#314) (74afd65)
• lac: add finite-horizon Lyapunov Candidate (#328) (ed2c85d)
• lac: implement Han et al. 2020 hyperparameters (#399) (574b651)
• latc: add LATC algorithm (#323) (d74c64f)
• package: renames the package to stable-learning-control (#257) (1133d0a)
• plot: improve plotter arguments (#425) (c7202a2)
• pytorch: add alpha/lambda learning rate customization (#412) (6feb749)
• replace OpenAi gym with Gymnasium (#255) (9873a03)
• tf2: add alpha/lambda learning rate customization (#416) (712e94b)
• wandb: add 'wandb_run_name' argument (#325) (e0a0b9d)

Bug Fixes

• Address flake8 and black formatting issues (#395) (517ee30)
• correct finite horizon buffer calculation (#398) (779201c)
• correctly close gymnasium environments (#340) (a179176)
• ensure 'test_policy' works with gymnasium>=0.28.1 (#276) (80fe370)
• fix 'test_policy' episode length bug (#292) (6d34f74)
• fix 'test_policy' rendering (#291) (48443ca)
• fix environment validation (#280) (a8a0346)
• fix plot dataframe loading (#338) (90e16e9)
• fix several env/policy load bugs (dddd4d8)
• fix several policy loading problems (51a664e)
• improve tensorflow lazy import (#272) (75192a4)
• pytorch: correct epoch-based learning rate decay behavior (#410) (a8df90f)
• pytorch: correct step-based learning rate decay (#405) (7d7ac76)
• pytorch: ensure correct application of constant learning rate (#411) (2b3693e)
• pytorch: fix learning rate decay defaults (#414) (27964fe)
• pytorch: resolve critical action rescaling bug (#403) (71d4f64)
• remove 'simzoo' submodule (#246) (0122aae)
• Resolve ruamel safe_load deprecation issue (#396) (cfcf81c)
• run: resolve issue with data_dir input argument (#409) (8d93610)
• tf2: correct off-by-one error in learning rate decay calculation (#415) (6ab5001)
• tf2: correct step-based learning rate decay (#407) (642a193)
• tf2: fix critical tf2 gradient update bug (#322) (dfc239b)
• torch: handle 'update_after' set to zero (#408) (7999590)
• wandb: fix wandb config format and run name (#317) (ca048de)

Documentation

• add docker ownership admonition (58a8e70)
• add policy test utility to docker page (5d39316)
• add pre-commit note (#341) (9853d05)
• add ros-gazebo-gym positive reward comment (#349) (9f5c434)
• add snapshots comment to CONTRIBUTING (c2031b2)
• add zenodo reference information (#300) (7180fb3)
• contributing: updates the contributing documentation (bac8787)
• fix broken external links (51317c0)
• fix contribution links (#245) (cccc5a7)
• fix doc code highlights (#404) (99126f7)
• fix documentation (#377) (c3b47c5)
• fix documentation latex formulas (#383) (#385) (582568b)
• fix incorrect package naming (#282) (5337841)
• fix incorrect run command (43df73f)
• fix links (#268) (21c5bcf)
• fix SAC alg run command (#380) (cf0a7a3)
• fix typos (30df6cb)
• improve code API documentation (#306) (16d5e26)
• improve docs (#344) (908fd78)
• improve documentation and package.json (#331) (e9fb90e)
• improve documentation defaults (be99008)
• improve documentation urls and syntax (#346) (22037cb)
• improve package description (#260) (eabfef3)
• improve polyak description (#417) (2404a4e)
• improve the documentation (#265) (829eb05)
• incorporate Docker usage instructions (#401) (2ba8031)
• sphinx: fixes documentation errors (38f2da0)
• sphinx: updates documentation installation instructions (ddadeb6)
• update admonitions to new GFM specs (ae7d3bc)
• update contribution note (#316) (44a35af)
• update docs make file (6eeb56e)
• update documentation (7ceee46)
• update documentation and docstrings (5548505)
• update numpy intersphinx link (581333d)
• update sandbox description (#329) (f60cc3b)
• updates code guidelines documentation (4f96a46)
• updates docs ci readme badge (074b369)

Code Refactoring

• buffers: remove unused 'rew_dim' argument (#327) (a69a7f6)