mirror of
https://github.com/saymrwulf/stable-baselines3.git
synced 2026-11-18 09:54:10 +00:00
Address minor issues after clarification by @araffin
This commit is contained in:
parent
e61d34a6f0
commit
91bbc28c0f
3 changed files with 23 additions and 4 deletions
|
|
@ -320,8 +320,8 @@ class BaseAlgorithm(ABC):
|
||||||
f"Stored kwargs: {data['policy_kwargs']}, specified kwargs: {kwargs['policy_kwargs']}")
|
f"Stored kwargs: {data['policy_kwargs']}, specified kwargs: {kwargs['policy_kwargs']}")
|
||||||
|
|
||||||
# check if observation space and action space are part of the saved parameters
|
# check if observation space and action space are part of the saved parameters
|
||||||
if ("observation_space" not in data or "action_space" not in data) and "env" not in data:
|
if "observation_space" not in data or "action_space" not in data:
|
||||||
raise ValueError("The observation_space and action_space was not given, can't verify new environments")
|
raise KeyError("The observation_space and action_space were not given, can't verify new environments")
|
||||||
# check if given env is valid
|
# check if given env is valid
|
||||||
if env is not None:
|
if env is not None:
|
||||||
check_for_correct_spaces(env, data["observation_space"], data["action_space"])
|
check_for_correct_spaces(env, data["observation_space"], data["action_space"])
|
||||||
|
|
@ -425,8 +425,10 @@ class BaseAlgorithm(ABC):
|
||||||
:return: (Tuple[int, BaseCallback])
|
:return: (Tuple[int, BaseCallback])
|
||||||
"""
|
"""
|
||||||
self.start_time = time.time()
|
self.start_time = time.time()
|
||||||
self.ep_info_buffer = deque(maxlen=100)
|
if self.ep_info_buffer is None or reset_num_timesteps:
|
||||||
self.ep_success_buffer = deque(maxlen=100)
|
# Initialize buffers if they don't exist, or reinitialize if resetting counters
|
||||||
|
self.ep_info_buffer = deque(maxlen=100)
|
||||||
|
self.ep_success_buffer = deque(maxlen=100)
|
||||||
|
|
||||||
if self.action_noise is not None:
|
if self.action_noise is not None:
|
||||||
self.action_noise.reset()
|
self.action_noise.reset()
|
||||||
|
|
|
||||||
|
|
@ -17,6 +17,20 @@ class Distribution(ABC):
|
||||||
def __init__(self):
|
def __init__(self):
|
||||||
super(Distribution, self).__init__()
|
super(Distribution, self).__init__()
|
||||||
|
|
||||||
|
@abstractmethod
|
||||||
|
def proba_distribution_net(self, *args, **kwargs):
|
||||||
|
"""Create the layers and parameters that represent the distribution.
|
||||||
|
|
||||||
|
Subclasses must define this, but the arguments and return type vary between
|
||||||
|
concrete classes."""
|
||||||
|
|
||||||
|
@abstractmethod
|
||||||
|
def proba_distribution(self, *args, **kwargs) -> 'Distribution':
|
||||||
|
"""Set parameters of the distribution.
|
||||||
|
|
||||||
|
:return: (Distribution) self
|
||||||
|
"""
|
||||||
|
|
||||||
@abstractmethod
|
@abstractmethod
|
||||||
def log_prob(self, x: th.Tensor) -> th.Tensor:
|
def log_prob(self, x: th.Tensor) -> th.Tensor:
|
||||||
"""
|
"""
|
||||||
|
|
|
||||||
|
|
@ -140,6 +140,7 @@ class BasePolicy(nn.Module, ABC):
|
||||||
:return: (Tuple[np.ndarray, Optional[np.ndarray]]) the model's action and the next state
|
:return: (Tuple[np.ndarray, Optional[np.ndarray]]) the model's action and the next state
|
||||||
(used in recurrent policies)
|
(used in recurrent policies)
|
||||||
"""
|
"""
|
||||||
|
# TODO (GH/1): add support for RNN policies
|
||||||
# if state is None:
|
# if state is None:
|
||||||
# state = self.initial_state
|
# state = self.initial_state
|
||||||
# if mask is None:
|
# if mask is None:
|
||||||
|
|
@ -438,6 +439,8 @@ class ActorCriticPolicy(BasePolicy):
|
||||||
self.action_net = self.action_dist.proba_distribution_net(latent_dim=latent_dim_pi)
|
self.action_net = self.action_dist.proba_distribution_net(latent_dim=latent_dim_pi)
|
||||||
elif isinstance(self.action_dist, BernoulliDistribution):
|
elif isinstance(self.action_dist, BernoulliDistribution):
|
||||||
self.action_net = self.action_dist.proba_distribution_net(latent_dim=latent_dim_pi)
|
self.action_net = self.action_dist.proba_distribution_net(latent_dim=latent_dim_pi)
|
||||||
|
else:
|
||||||
|
raise NotImplementedError(f"Unsupported distribution '{self.action_dist}'.")
|
||||||
|
|
||||||
self.value_net = nn.Linear(self.mlp_extractor.latent_dim_vf, 1)
|
self.value_net = nn.Linear(self.mlp_extractor.latent_dim_vf, 1)
|
||||||
# Init weights: use orthogonal initialization
|
# Init weights: use orthogonal initialization
|
||||||
|
|
|
||||||
Loading…
Reference in a new issue