Releases
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
Code Refactoring
buffers: remove unused 'rew_dim' argument (#327 ) (a69a7f6 )
You can’t perform that action at this time.