stable-baselines3/stable_baselines3/common/vec_env/__init__.py

106 lines
4.3 KiB
Python
Raw Normal View History

2019-11-14 13:35:00 +00:00
from copy import deepcopy
from typing import Optional, TypeVar
from stable_baselines3.common.vec_env.base_vec_env import CloudpickleWrapper, VecEnv, VecEnvWrapper
2020-05-05 13:02:35 +00:00
from stable_baselines3.common.vec_env.dummy_vec_env import DummyVecEnv
from stable_baselines3.common.vec_env.stacked_observations import StackedObservations
2020-05-05 13:02:35 +00:00
from stable_baselines3.common.vec_env.subproc_vec_env import SubprocVecEnv
from stable_baselines3.common.vec_env.vec_check_nan import VecCheckNan
from stable_baselines3.common.vec_env.vec_extract_dict_obs import VecExtractDictObs
2020-05-05 13:02:35 +00:00
from stable_baselines3.common.vec_env.vec_frame_stack import VecFrameStack
from stable_baselines3.common.vec_env.vec_monitor import VecMonitor
2020-05-05 13:02:35 +00:00
from stable_baselines3.common.vec_env.vec_normalize import VecNormalize
from stable_baselines3.common.vec_env.vec_transpose import VecTransposeImage
2020-05-05 14:28:38 +00:00
from stable_baselines3.common.vec_env.vec_video_recorder import VecVideoRecorder
2019-11-14 13:35:00 +00:00
VecEnvWrapperT = TypeVar("VecEnvWrapperT", bound=VecEnvWrapper)
2019-11-14 13:35:00 +00:00
2020-03-12 11:34:25 +00:00
def unwrap_vec_wrapper(env: VecEnv, vec_wrapper_class: type[VecEnvWrapperT]) -> Optional[VecEnvWrapperT]:
2019-11-14 13:35:00 +00:00
"""
Retrieve a ``VecEnvWrapper`` object by recursively searching.
:param env: The ``VecEnv`` that is going to be unwrapped
:param vec_wrapper_class: The desired ``VecEnvWrapper`` class.
:return: The ``VecEnvWrapper`` object if the ``VecEnv`` is wrapped with the desired wrapper, None otherwise
2019-11-14 13:35:00 +00:00
"""
env_tmp = env
while isinstance(env_tmp, VecEnvWrapper):
if isinstance(env_tmp, vec_wrapper_class):
2019-11-14 13:35:00 +00:00
return env_tmp
env_tmp = env_tmp.venv
return None
def unwrap_vec_normalize(env: VecEnv) -> Optional[VecNormalize]:
"""
Retrieve a ``VecNormalize`` object by recursively searching.
:param env: The VecEnv that is going to be unwrapped
:return: The ``VecNormalize`` object if the ``VecEnv`` is wrapped with ``VecNormalize``, None otherwise
"""
return unwrap_vec_wrapper(env, VecNormalize)
def is_vecenv_wrapped(env: VecEnv, vec_wrapper_class: type[VecEnvWrapper]) -> bool:
Implement HER (#120) * Added working her version, Online sampling is missing. * Updated test_her. * Added first version of online her sampling. Still problems with tensor dimensions. * Reformat * Fixed tests * Added some comments. * Updated changelog. * Add missing init file * Fixed some small bugs. * Reduced arguments for HER, small changes. * Added getattr. Fixed bug for online sampling. * Updated save/load funtions. Small changes. * Added her to init. * Updated save method. * Updated her ratio. * Move obs_wrapper * Added DQN test. * Fix potential bug * Offline and online her share same sample_goal function. * Changed lists into arrays. * Updated her test. * Fix online sampling * Fixed action bug. Updated time limit for episodes. * Updated convert_dict method to take keys as arguments. * Renamed obs dict wrapper. * Seed bit flipping env * Remove get_episode_dict * Add fast online sampling version * Added documentation. * Vectorized reward computation * Vectorized goal sampling * Update time limit for episodes in online her sampling. * Fix max episode length inference * Bug fix for Fetch envs * Fix for HER + gSDE * Reformat (new black version) * Added info dict to compute new reward. Check her_replay_buffer again. * Fix info buffer * Updated done flag. * Fixes for gSDE * Offline her version uses now HerReplayBuffer as episode storage. * Fix num_timesteps computation * Fix get torch params * Vectorized version for offline sampling. * Modified offline her sampling to use sample method of her_replay_buffer * Updated HER tests. * Updated documentation * Cleanup docstrings * Updated to review comments * Fix pytype * Update according to review comments. * Removed random goal strategy. Updated sample transitions. * Updated migration. Removed time signal removal. * Update doc * Fix potential load issue * Add VecNormalize support for dict obs * Updated saving/loading replay buffer for HER. * Fix test memory usage * Fixed save/load replay buffer. * Fixed save/load replay buffer * Fixed transition index after loading replay buffer in online sampling * Better error handling * Add tests for get_time_limit * More tests for VecNormalize with dict obs * Update doc * Improve HER description * Add test for sde support * Add comments * Add comments * Remove check that was always valid * Fix for terminal observation * Updated buffer size in offline version and reset of HER buffer * Reformat * Update doc * Remove np.empty + add doc * Fix loading * Updated loading replay buffer * Separate online and offline sampling + bug fixes * Update tensorboard log name * Version bump * Bug fix for special case Co-authored-by: Antonin Raffin <antonin.raffin@dlr.de> Co-authored-by: Antonin RAFFIN <antonin.raffin@ensta.org>
2020-10-22 09:56:43 +00:00
"""
Check if an environment is already wrapped in a given ``VecEnvWrapper``.
Implement HER (#120) * Added working her version, Online sampling is missing. * Updated test_her. * Added first version of online her sampling. Still problems with tensor dimensions. * Reformat * Fixed tests * Added some comments. * Updated changelog. * Add missing init file * Fixed some small bugs. * Reduced arguments for HER, small changes. * Added getattr. Fixed bug for online sampling. * Updated save/load funtions. Small changes. * Added her to init. * Updated save method. * Updated her ratio. * Move obs_wrapper * Added DQN test. * Fix potential bug * Offline and online her share same sample_goal function. * Changed lists into arrays. * Updated her test. * Fix online sampling * Fixed action bug. Updated time limit for episodes. * Updated convert_dict method to take keys as arguments. * Renamed obs dict wrapper. * Seed bit flipping env * Remove get_episode_dict * Add fast online sampling version * Added documentation. * Vectorized reward computation * Vectorized goal sampling * Update time limit for episodes in online her sampling. * Fix max episode length inference * Bug fix for Fetch envs * Fix for HER + gSDE * Reformat (new black version) * Added info dict to compute new reward. Check her_replay_buffer again. * Fix info buffer * Updated done flag. * Fixes for gSDE * Offline her version uses now HerReplayBuffer as episode storage. * Fix num_timesteps computation * Fix get torch params * Vectorized version for offline sampling. * Modified offline her sampling to use sample method of her_replay_buffer * Updated HER tests. * Updated documentation * Cleanup docstrings * Updated to review comments * Fix pytype * Update according to review comments. * Removed random goal strategy. Updated sample transitions. * Updated migration. Removed time signal removal. * Update doc * Fix potential load issue * Add VecNormalize support for dict obs * Updated saving/loading replay buffer for HER. * Fix test memory usage * Fixed save/load replay buffer. * Fixed save/load replay buffer * Fixed transition index after loading replay buffer in online sampling * Better error handling * Add tests for get_time_limit * More tests for VecNormalize with dict obs * Update doc * Improve HER description * Add test for sde support * Add comments * Add comments * Remove check that was always valid * Fix for terminal observation * Updated buffer size in offline version and reset of HER buffer * Reformat * Update doc * Remove np.empty + add doc * Fix loading * Updated loading replay buffer * Separate online and offline sampling + bug fixes * Update tensorboard log name * Version bump * Bug fix for special case Co-authored-by: Antonin Raffin <antonin.raffin@dlr.de> Co-authored-by: Antonin RAFFIN <antonin.raffin@ensta.org>
2020-10-22 09:56:43 +00:00
:param env: The VecEnv that is going to be checked
:param vec_wrapper_class: The desired ``VecEnvWrapper`` class.
:return: True if the ``VecEnv`` is wrapped with the desired wrapper, False otherwise
Implement HER (#120) * Added working her version, Online sampling is missing. * Updated test_her. * Added first version of online her sampling. Still problems with tensor dimensions. * Reformat * Fixed tests * Added some comments. * Updated changelog. * Add missing init file * Fixed some small bugs. * Reduced arguments for HER, small changes. * Added getattr. Fixed bug for online sampling. * Updated save/load funtions. Small changes. * Added her to init. * Updated save method. * Updated her ratio. * Move obs_wrapper * Added DQN test. * Fix potential bug * Offline and online her share same sample_goal function. * Changed lists into arrays. * Updated her test. * Fix online sampling * Fixed action bug. Updated time limit for episodes. * Updated convert_dict method to take keys as arguments. * Renamed obs dict wrapper. * Seed bit flipping env * Remove get_episode_dict * Add fast online sampling version * Added documentation. * Vectorized reward computation * Vectorized goal sampling * Update time limit for episodes in online her sampling. * Fix max episode length inference * Bug fix for Fetch envs * Fix for HER + gSDE * Reformat (new black version) * Added info dict to compute new reward. Check her_replay_buffer again. * Fix info buffer * Updated done flag. * Fixes for gSDE * Offline her version uses now HerReplayBuffer as episode storage. * Fix num_timesteps computation * Fix get torch params * Vectorized version for offline sampling. * Modified offline her sampling to use sample method of her_replay_buffer * Updated HER tests. * Updated documentation * Cleanup docstrings * Updated to review comments * Fix pytype * Update according to review comments. * Removed random goal strategy. Updated sample transitions. * Updated migration. Removed time signal removal. * Update doc * Fix potential load issue * Add VecNormalize support for dict obs * Updated saving/loading replay buffer for HER. * Fix test memory usage * Fixed save/load replay buffer. * Fixed save/load replay buffer * Fixed transition index after loading replay buffer in online sampling * Better error handling * Add tests for get_time_limit * More tests for VecNormalize with dict obs * Update doc * Improve HER description * Add test for sde support * Add comments * Add comments * Remove check that was always valid * Fix for terminal observation * Updated buffer size in offline version and reset of HER buffer * Reformat * Update doc * Remove np.empty + add doc * Fix loading * Updated loading replay buffer * Separate online and offline sampling + bug fixes * Update tensorboard log name * Version bump * Bug fix for special case Co-authored-by: Antonin Raffin <antonin.raffin@dlr.de> Co-authored-by: Antonin RAFFIN <antonin.raffin@ensta.org>
2020-10-22 09:56:43 +00:00
"""
return unwrap_vec_wrapper(env, vec_wrapper_class) is not None
def sync_envs_normalization(env: VecEnv, eval_env: VecEnv) -> None:
2019-11-14 13:35:00 +00:00
"""
Synchronize the normalization statistics of an eval environment and train environment
when they are both wrapped in a ``VecNormalize`` wrapper.
2019-11-14 13:35:00 +00:00
:param env: Training env
:param eval_env: Environment used for evaluation.
2019-11-14 13:35:00 +00:00
"""
env_tmp, eval_env_tmp = env, eval_env
while isinstance(env_tmp, VecEnvWrapper):
assert isinstance(eval_env_tmp, VecEnvWrapper), (
"Error while synchronizing normalization stats: expected the eval env to be "
f"a VecEnvWrapper but got {eval_env_tmp} instead. "
"This is probably due to the training env not being wrapped the same way as the evaluation env. "
f"Training env type: {env_tmp}."
)
2019-11-14 13:35:00 +00:00
if isinstance(env_tmp, VecNormalize):
assert isinstance(eval_env_tmp, VecNormalize), (
"Error while synchronizing normalization stats: expected the eval env to be "
f"a VecNormalize but got {eval_env_tmp} instead. "
"This is probably due to the training env not being wrapped the same way as the evaluation env. "
f"Training env type: {env_tmp}."
)
# Only synchronize if observation normalization exists
if hasattr(env_tmp, "obs_rms"):
eval_env_tmp.obs_rms = deepcopy(env_tmp.obs_rms)
2020-03-12 11:34:25 +00:00
eval_env_tmp.ret_rms = deepcopy(env_tmp.ret_rms)
2019-11-14 13:35:00 +00:00
env_tmp = env_tmp.venv
2020-01-20 10:17:55 +00:00
eval_env_tmp = eval_env_tmp.venv
__all__ = [
"CloudpickleWrapper",
"DummyVecEnv",
"StackedObservations",
"SubprocVecEnv",
"VecCheckNan",
"VecEnv",
"VecEnvWrapper",
"VecExtractDictObs",
"VecFrameStack",
"VecMonitor",
"VecNormalize",
"VecTransposeImage",
"VecVideoRecorder",
"is_vecenv_wrapped",
"sync_envs_normalization",
"unwrap_vec_normalize",
"unwrap_vec_wrapper",
]