# Copyright 2021 Sony Corporation.
# Copyright 2021,2022,2023 Sony Group Corporation.
#
# Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.
import random
from dataclasses import dataclass
from typing import Any, Dict, Optional, Union
import gym
import numpy as np
import nnabla as nn
import nnabla.functions as NF
import nnabla.solvers as NS
import nnabla_rl.environment_explorers as EE
import nnabla_rl.model_trainers as MT
import nnabla_rl.preprocessors as RP
from nnabla_rl.algorithm import Algorithm, AlgorithmConfig, eval_api
from nnabla_rl.algorithms.common_utils import (_StatePreprocessedRewardFunction, _StatePreprocessedStochasticPolicy,
_StatePreprocessedVFunction, _StochasticPolicyActionSelector,
compute_v_target_and_advantage)
from nnabla_rl.builders import ExplorerBuilder, ModelBuilder, PreprocessorBuilder, SolverBuilder
from nnabla_rl.environment_explorer import EnvironmentExplorer
from nnabla_rl.environments.environment_info import EnvironmentInfo
from nnabla_rl.model_trainers.model_trainer import ModelTrainer, TrainingBatch
from nnabla_rl.models import (GAILDiscriminator, GAILPolicy, GAILVFunction, Model, RewardFunction, StochasticPolicy,
VFunction)
from nnabla_rl.preprocessors import Preprocessor
from nnabla_rl.replay_buffer import ReplayBuffer
from nnabla_rl.replay_buffers.buffer_iterator import BufferIterator
from nnabla_rl.utils import context
from nnabla_rl.utils.data import add_batch_dimension, marshal_experiences, set_data_to_variable
from nnabla_rl.utils.misc import create_variable
[docs]@dataclass
class GAILConfig(AlgorithmConfig):
"""List of configurations for GAIL algorithm.
Args:
act_deterministic_in_eval (bool): Enable act deterministically at evalution. Defaults to True.
discriminator_batch_size (bool): Trainig batch size of discriminator.\
Usually, discriminator_batch_size is the same as pi_batch_size. Defaults to 50000.
discriminator_learning_rate (float): Learning rate which is set to the solvers of dicriminator function. \
You can customize/override the learning rate for each solver by implementing the \
(:py:class:`SolverBuilder <nnabla_rl.builders.SolverBuilder>`) by yourself. \
Defaults to 0.001.
discriminator_update_frequency (int): Frequency (measured in the number of parameter update) \
of discriminator update. Defaults to 1.
adversary_entropy_coef (float): Coefficient of entropy loss in dicriminator training. Defaults to 0.001.
policy_update_frequency (int): Frequency (measured in the number of parameter update) \
of policy update. Defaults to 1.
gamma (float): Discount factor of rewards. Defaults to 0.995.
lmb (float): Scalar of lambda return's computation in GAE. Defaults to 0.97.\
This configuration is related to bias and variance of estimated value. \
If it is close to 0, estimated value is low-variance but biased.\
If it is close to 1, estimated value is unbiased but high-variance.
num_steps_per_iteration (int): Number of steps per each training iteration for collecting on-policy experinces.\
Increasing this step size is effective to get precise parameters of policy and value function updating, \
but computational time of each iteration will increase. Defaults to 50000.
pi_batch_size (int): Trainig batch size of policy. \
Usually, pi_batch_size is the same as num_steps_per_iteration. Defaults to 50000.
sigma_kl_divergence_constraint (float): Constraint size of kl divergence \
between previous policy and updated policy. Defaults to 0.01.
maximum_backtrack_numbers (int): Maximum backtrack numbers of linesearch. Defaults to 10.
conjugate_gradient_damping (float): Damping size of conjugate gradient method. Defaults to 0.1.
conjugate_gradient_iterations (int): Number of iterations of conjugate gradient method. Defaults to 10.
vf_epochs (int): Number of epochs in each iteration. Defaults to 5.
vf_batch_size (int): Training batch size of value function. Defaults to 128.
vf_learning_rate (float): Learning rate which is set to the solvers of value function. \
You can customize/override the learning rate for each solver by implementing the \
(:py:class:`SolverBuilder <nnabla_rl.builders.SolverBuilder>`) by yourself. \
Defaults to 0.001.
preprocess_state (bool): Enable preprocessing the states in the collected experiences \
before feeding as training batch. Defaults to True.
"""
# type declarations to type check with mypy
# NOTE: declared variables are instance variable and NOT class variable, unless it is marked with ClassVar
# See https://mypy.readthedocs.io/en/stable/class_basics.html for details
preprocess_state: bool = True
act_deterministic_in_eval: bool = True
discriminator_batch_size: int = 50000
discriminator_learning_rate: float = 0.01
discriminator_update_frequency: int = 1
adversary_entropy_coef: float = 0.001
policy_update_frequency: int = 1
gamma: float = 0.995
lmb: float = 0.97
pi_batch_size: int = 50000
num_steps_per_iteration: int = 50000
sigma_kl_divergence_constraint: float = 0.01
maximum_backtrack_numbers: int = 10
conjugate_gradient_damping: float = 0.1
conjugate_gradient_iterations: int = 10
vf_epochs: int = 5
vf_batch_size: int = 128
vf_learning_rate: float = 1e-3
def __post_init__(self):
"""__post_init__
Check the values are in valid range.
"""
self._assert_between(self.pi_batch_size, 0, self.num_steps_per_iteration, 'pi_batch_size')
self._assert_positive(self.discriminator_learning_rate, "discriminator_learning_rate")
self._assert_positive(self.discriminator_batch_size, "discriminator_batch_size")
self._assert_positive(self.policy_update_frequency, "policy_update_frequency")
self._assert_positive(self.discriminator_update_frequency, "discriminator_update_frequency")
self._assert_positive(self.adversary_entropy_coef, "adversarial_entropy_coef")
self._assert_between(self.gamma, 0.0, 1.0, 'gamma')
self._assert_between(self.lmb, 0.0, 1.0, 'lmb')
self._assert_positive(self.num_steps_per_iteration, 'num_steps_per_iteration')
self._assert_positive(self.sigma_kl_divergence_constraint, 'sigma_kl_divergence_constraint')
self._assert_positive(self.maximum_backtrack_numbers, 'maximum_backtrack_numbers')
self._assert_positive(self.conjugate_gradient_damping, 'conjugate_gradient_damping')
self._assert_positive(self.conjugate_gradient_iterations, 'conjugate_gradient_iterations')
self._assert_positive(self.vf_epochs, 'vf_epochs')
self._assert_positive(self.vf_batch_size, 'vf_batch_size')
self._assert_positive(self.vf_learning_rate, 'vf_learning_rate')
class DefaultPreprocessorBuilder(PreprocessorBuilder):
def build_preprocessor(self, # type: ignore[override]
scope_name: str,
env_info: EnvironmentInfo,
algorithm_config: GAILConfig,
**kwargs) -> Preprocessor:
return RP.RunningMeanNormalizer(scope_name, env_info.state_shape, value_clip=(-5.0, 5.0))
class DefaultPolicyBuilder(ModelBuilder[StochasticPolicy]):
def build_model(self, # type: ignore[override]
scope_name: str,
env_info: EnvironmentInfo,
algorithm_config: GAILConfig,
**kwargs) -> StochasticPolicy:
return GAILPolicy(scope_name, env_info.action_dim)
class DefaultVFunctionBuilder(ModelBuilder[VFunction]):
def build_model(self, # type: ignore[override]
scope_name: str,
env_info: EnvironmentInfo,
algorithm_config: GAILConfig,
**kwargs) -> VFunction:
return GAILVFunction(scope_name)
class DefaultRewardFunctionBuilder(ModelBuilder[RewardFunction]):
def build_model(self, # type: ignore[override]
scope_name: str,
env_info: EnvironmentInfo,
algorithm_config: GAILConfig,
**kwargs) -> RewardFunction:
return GAILDiscriminator(scope_name)
class DefaultVFunctionSolverBuilder(SolverBuilder):
def build_solver(self, # type: ignore[override]
env_info: EnvironmentInfo,
algorithm_config: GAILConfig,
**kwargs) -> nn.solver.Solver:
return NS.Adam(alpha=algorithm_config.vf_learning_rate)
class DefaultRewardFunctionSolverBuilder(SolverBuilder):
def build_solver(self, # type: ignore[override]
env_info: EnvironmentInfo,
algorithm_config: GAILConfig,
**kwargs) -> nn.solver.Solver:
assert isinstance(algorithm_config, GAILConfig)
return NS.Adam(alpha=algorithm_config.discriminator_learning_rate)
class DefaultExplorerBuilder(ExplorerBuilder):
def build_explorer(self, # type: ignore[override]
env_info: EnvironmentInfo,
algorithm_config: GAILConfig,
algorithm: "GAIL",
**kwargs) -> EnvironmentExplorer:
explorer_config = EE.RawPolicyExplorerConfig(
initial_step_num=algorithm.iteration_num,
timelimit_as_terminal=False
)
explorer = EE.RawPolicyExplorer(policy_action_selector=algorithm._exploration_action_selector,
env_info=env_info,
config=explorer_config)
return explorer
[docs]class GAIL(Algorithm):
"""Generative Adversarial Imitation Learning implementation.
This class implements the Generative Adversarial Imitation Learning (GAIL) algorithm
proposed by Jonathan Ho, et al. in the paper: "Generative Adversarial Imitation Learning"
For detail see: https://arxiv.org/abs/1606.03476
This algorithm only supports online training.
Args:
env_or_env_info\
(gym.Env or :py:class:`EnvironmentInfo <nnabla_rl.environments.environment_info.EnvironmentInfo>`):
the environment to train or environment info
expert_buffer (:py:class:`ReplayBuffer <nnabla_rl.replay_buffer.ReplayBuffer>`):
replay buffer which contains expert experience.
config (:py:class:`GAILConfig <nnabla_rl.algorithms.gail.GAILConfig>`): configuration of GAIL algorithm
v_function_builder (:py:class:`ModelBuilder[VFunction] <nnabla_rl.builders.ModelBuilder>`):
builder of v function models
v_solver_builder (:py:class:`SolverBuilder <nnabla_rl.builders.SolverBuilder>`): builder for v function solvers
policy_builder (:py:class:`ModelBuilder[StochasicPolicy] <nnabla_rl.builders.ModelBuilder>`):
builder of policy models
reward_function_builder (:py:class:`ModelBuilder[RewardFunction] <nnabla_rl.builders.ModelBuilder>`):
builder of reward function models
reward_solver_builder (:py:class:`SolverBuilder <nnabla_rl.builders.SolverBuilder>`):
builder for reward function solvers
state_preprocessor_builder (None or :py:class:`PreprocessorBuilder <nnabla_rl.builders.PreprocessorBuilder>`):
state preprocessor builder to preprocess the states
explorer_builder (:py:class:`ExplorerBuilder <nnabla_rl.builders.ExplorerBuilder>`):
builder of environment explorer
"""
_config: GAILConfig
_v_function: VFunction
_v_function_solver: nn.solver.Solver
_policy: StochasticPolicy
_discriminator: RewardFunction
_discriminator_solver: nn.solver.Solver
_explorer_builder: ExplorerBuilder
_environment_explorer: EnvironmentExplorer
_v_function_trainer: ModelTrainer
_policy_trainer: ModelTrainer
_discriminator_trainer: ModelTrainer
_s_var_label: nn.Variable
_s_next_var_label: nn.Variable
_a_var_label: nn.Variable
_reward: nn.Variable
_v_function_trainer_state: Dict[str, Any]
_policy_trainer_state: Dict[str, Any]
_discriminator_trainer_state: Dict[str, Any]
_evaluation_actor: _StochasticPolicyActionSelector
_exploration_actor: _StochasticPolicyActionSelector
def __init__(self, env_or_env_info: Union[gym.Env, EnvironmentInfo],
expert_buffer: ReplayBuffer,
config: GAILConfig = GAILConfig(),
v_function_builder: ModelBuilder[VFunction] = DefaultVFunctionBuilder(),
v_solver_builder: SolverBuilder = DefaultVFunctionSolverBuilder(),
policy_builder: ModelBuilder[StochasticPolicy] = DefaultPolicyBuilder(),
reward_function_builder: ModelBuilder[RewardFunction] = DefaultRewardFunctionBuilder(),
reward_solver_builder: SolverBuilder = DefaultRewardFunctionSolverBuilder(),
state_preprocessor_builder: Optional[PreprocessorBuilder] = DefaultPreprocessorBuilder(),
explorer_builder: ExplorerBuilder = DefaultExplorerBuilder()):
super(GAIL, self).__init__(env_or_env_info, config=config)
self._explorer_builder = explorer_builder
with nn.context_scope(context.get_nnabla_context(self._config.gpu_id)):
policy = policy_builder("pi", self._env_info, self._config)
v_function = v_function_builder("v", self._env_info, self._config)
discriminator = reward_function_builder("discriminator", self._env_info, self._config)
if self._config.preprocess_state:
if state_preprocessor_builder is None:
raise ValueError('State preprocessing is enabled but no preprocessor builder is given')
pi_v_preprocessor = state_preprocessor_builder('pi_v_preprocessor', self._env_info, self._config)
v_function = _StatePreprocessedVFunction(v_function=v_function, preprocessor=pi_v_preprocessor)
policy = _StatePreprocessedStochasticPolicy(policy=policy, preprocessor=pi_v_preprocessor)
r_preprocessor = state_preprocessor_builder('r_preprocessor', self._env_info, self._config)
discriminator = _StatePreprocessedRewardFunction(
reward_function=discriminator, preprocessor=r_preprocessor)
self._pi_v_state_preprocessor = pi_v_preprocessor
self._r_state_preprocessor = r_preprocessor
self._v_function = v_function
self._policy = policy
self._discriminator = discriminator
self._v_function_solver = v_solver_builder(self._env_info, self._config)
self._discriminator_solver = reward_solver_builder(self._env_info, self._config)
self._expert_buffer = expert_buffer
self._evaluation_actor = _StochasticPolicyActionSelector(
self._env_info, self._policy.shallowcopy(), deterministic=self._config.act_deterministic_in_eval)
self._exploration_actor = _StochasticPolicyActionSelector(
self._env_info, self._policy.shallowcopy(), deterministic=False)
@eval_api
def compute_eval_action(self, state, *, begin_of_episode=False, extra_info={}):
with nn.context_scope(context.get_nnabla_context(self._config.gpu_id)):
action, _ = self._evaluation_action_selector(state, begin_of_episode=begin_of_episode)
return action
def _before_training_start(self, env_or_buffer):
# set context globally to ensure that the training runs on configured gpu
context.set_nnabla_context(self._config.gpu_id)
self._environment_explorer = self._setup_environment_explorer(env_or_buffer)
self._v_function_trainer = self._setup_v_function_training(env_or_buffer)
self._policy_trainer = self._setup_policy_training(env_or_buffer)
self._discriminator_trainer = self._setup_reward_function_training(env_or_buffer)
def _setup_environment_explorer(self, env_or_buffer):
return None if self._is_buffer(env_or_buffer) else self._explorer_builder(self._env_info, self._config, self)
def _setup_v_function_training(self, env_or_buffer):
v_function_trainer_config = MT.v_value_trainers.MonteCarloVTrainerConfig(
reduction_method='mean',
v_loss_scalar=1.0
)
v_function_trainer = MT.v_value_trainers.MonteCarloVTrainer(
train_functions=self._v_function,
solvers={self._v_function.scope_name: self._v_function_solver},
env_info=self._env_info,
config=v_function_trainer_config)
return v_function_trainer
def _setup_policy_training(self, env_or_buffer):
policy_trainer_config = MT.policy_trainers.TRPOPolicyTrainerConfig(
sigma_kl_divergence_constraint=self._config.sigma_kl_divergence_constraint,
maximum_backtrack_numbers=self._config.maximum_backtrack_numbers,
conjugate_gradient_damping=self._config.conjugate_gradient_damping,
conjugate_gradient_iterations=self._config.conjugate_gradient_iterations)
policy_trainer = MT.policy_trainers.TRPOPolicyTrainer(
model=self._policy,
env_info=self._env_info,
config=policy_trainer_config)
return policy_trainer
def _setup_reward_function_training(self, env_or_buffer):
reward_function_trainer_config = MT.reward_trainiers.GAILRewardFunctionTrainerConfig(
batch_size=self._config.discriminator_batch_size,
learning_rate=self._config.discriminator_learning_rate,
entropy_coef=self._config.adversary_entropy_coef
)
reward_function_trainer = MT.reward_trainiers.GAILRewardFunctionTrainer(
models=self._discriminator,
solvers={self._discriminator.scope_name: self._discriminator_solver},
env_info=self._env_info,
config=reward_function_trainer_config)
return reward_function_trainer
def _run_online_training_iteration(self, env):
if self.iteration_num % self._config.num_steps_per_iteration != 0:
return
buffer = ReplayBuffer(capacity=self._config.num_steps_per_iteration)
num_steps = 0
while num_steps <= self._config.num_steps_per_iteration:
experience = self._environment_explorer.rollout(env)
experience = self._label_experience(experience)
buffer.append(experience)
num_steps += len(experience)
self._gail_training(buffer)
def _label_experience(self, experience):
labeled_experience = []
if not hasattr(self, '_s_var_label'):
# build graph
self._s_var_label = create_variable(1, self._env_info.state_shape)
self._s_next_var_label = create_variable(1, self._env_info.state_shape)
self._a_var_label = create_variable(1, self._env_info.action_shape)
logits_fake = self._discriminator.r(self._s_var_label, self._a_var_label, self._s_next_var_label)
self._reward = -NF.log(1. - NF.sigmoid(logits_fake) + 1e-8)
for s, a, _, non_terminal, n_s, info in experience:
# forward and get reward
set_data_to_variable(self._s_var_label, add_batch_dimension(s))
set_data_to_variable(self._a_var_label, add_batch_dimension(a))
set_data_to_variable(self._s_next_var_label, add_batch_dimension(n_s))
self._reward.forward()
transition = (s, a, self._reward.d, non_terminal, n_s, info)
labeled_experience.append(transition)
return labeled_experience
def _run_offline_training_iteration(self, buffer):
raise NotImplementedError
def _gail_training(self, buffer):
buffer_iterator = BufferIterator(buffer, 1, shuffle=False, repeat=False)
# policy learning
if self._iteration_num % self._config.policy_update_frequency == 0:
s, a, v_target, advantage = self._align_policy_experiences(buffer_iterator)
if self._config.preprocess_state:
self._pi_v_state_preprocessor.update(s)
self._policy_training(s, a, v_target, advantage)
self._v_function_training(s, v_target)
# discriminator learning
if self._iteration_num % self._config.discriminator_update_frequency == 0:
s_curr_expert, a_curr_expert, s_next_expert, s_curr_agent, a_curr_agent, s_next_agent = \
self._align_discriminator_experiences(buffer_iterator)
if self._config.preprocess_state:
self._r_state_preprocessor.update(np.concatenate([s_curr_agent, s_curr_expert], axis=0))
self._discriminator_training(s_curr_expert, a_curr_expert, s_next_expert,
s_curr_agent, a_curr_agent, s_next_agent)
def _align_policy_experiences(self, buffer_iterator):
v_target_batch, adv_batch = self._compute_v_target_and_advantage(buffer_iterator)
s_batch, a_batch, _ = self._align_state_and_action(buffer_iterator)
return s_batch[:self._config.num_steps_per_iteration], \
a_batch[:self._config.num_steps_per_iteration], \
v_target_batch[:self._config.num_steps_per_iteration], \
adv_batch[:self._config.num_steps_per_iteration]
def _compute_v_target_and_advantage(self, buffer_iterator):
v_target_batch = []
adv_batch = []
buffer_iterator.reset()
for experiences, *_ in buffer_iterator:
# length of experiences is 1
v_target, adv = compute_v_target_and_advantage(
self._v_function, experiences[0], gamma=self._config.gamma, lmb=self._config.lmb)
v_target_batch.append(v_target.reshape(-1, 1))
adv_batch.append(adv.reshape(-1, 1))
adv_batch = np.concatenate(adv_batch, axis=0)
v_target_batch = np.concatenate(v_target_batch, axis=0)
adv_mean = np.mean(adv_batch)
adv_std = np.std(adv_batch)
adv_batch = (adv_batch - adv_mean) / adv_std
return v_target_batch, adv_batch
def _align_state_and_action(self, buffer_iterator, batch_size=None):
s_batch = []
a_batch = []
s_next_batch = []
buffer_iterator.reset()
for experiences, _ in buffer_iterator:
# length of experiences is 1
s_seq, a_seq, _, _, s_next_seq, *_ = marshal_experiences(experiences[0])
s_batch.append(s_seq)
a_batch.append(a_seq)
s_next_batch.append(s_next_seq)
s_batch = np.concatenate(s_batch, axis=0)
a_batch = np.concatenate(a_batch, axis=0)
s_next_batch = np.concatenate(s_next_batch, axis=0)
if batch_size is None:
return s_batch, a_batch, s_next_batch
idx = random.sample(list(range(s_batch.shape[0])), batch_size)
return s_batch[idx], a_batch[idx], s_next_batch[idx]
def _align_discriminator_experiences(self, buffer_iterator):
# sample expert data
expert_experience, _ = self._expert_buffer.sample(self._config.discriminator_batch_size)
s_expert_batch, a_expert_batch, _, _, s_next_expert_batch, *_ = marshal_experiences(expert_experience)
# sample agent data
s_batch, a_batch, s_next_batch = self._align_state_and_action(
buffer_iterator, batch_size=self._config.discriminator_batch_size)
return s_expert_batch, a_expert_batch, s_next_expert_batch, s_batch, a_batch, s_next_batch
def _v_function_training(self, s, v_target):
num_iterations_per_epoch = self._config.num_steps_per_iteration // self._config.vf_batch_size
for _ in range(self._config.vf_epochs * num_iterations_per_epoch):
indices = np.random.randint(0, self._config.num_steps_per_iteration, size=self._config.vf_batch_size)
batch = TrainingBatch(batch_size=self._config.vf_batch_size,
s_current=s[indices],
extra={'v_target': v_target[indices]})
self._v_function_trainer_state = self._v_function_trainer.train(batch)
def _policy_training(self, s, a, v_target, advantage):
extra = {}
extra['v_target'] = v_target[:self._config.pi_batch_size]
extra['advantage'] = advantage[:self._config.pi_batch_size]
batch = TrainingBatch(batch_size=self._config.pi_batch_size,
s_current=s[:self._config.pi_batch_size],
a_current=a[:self._config.pi_batch_size],
extra=extra)
self._policy_trainer_state = self._policy_trainer.train(batch)
def _discriminator_training(self, s_curr_expert, a_curr_expert, s_next_expert,
s_curr_agent, a_curr_agent, s_next_agent):
extra = {}
extra['s_current_agent'] = s_curr_agent[:self._config.discriminator_batch_size]
extra['a_current_agent'] = a_curr_agent[:self._config.discriminator_batch_size]
extra['s_next_agent'] = s_next_agent[:self._config.discriminator_batch_size]
extra['s_current_expert'] = s_curr_expert[:self._config.discriminator_batch_size]
extra['a_current_expert'] = a_curr_expert[:self._config.discriminator_batch_size]
extra['s_next_expert'] = s_next_expert[:self._config.discriminator_batch_size]
batch = TrainingBatch(batch_size=self._config.discriminator_batch_size,
extra=extra)
self._discriminator_trainer_state = self._discriminator_trainer.train(batch)
def _evaluation_action_selector(self, s, *, begin_of_episode=False):
return self._evaluation_actor(s, begin_of_episode=begin_of_episode)
def _exploration_action_selector(self, s, *, begin_of_episode=False):
return self._exploration_actor(s, begin_of_episode=begin_of_episode)
def _models(self):
models = {}
models[self._policy.scope_name] = self._policy
models[self._v_function.scope_name] = self._v_function
models[self._discriminator.scope_name] = self._discriminator
if self._config.preprocess_state and isinstance(self._r_state_preprocessor, Model):
models[self._r_state_preprocessor.scope_name] = self._r_state_preprocessor
if self._config.preprocess_state and isinstance(self._pi_v_state_preprocessor, Model):
models[self._pi_v_state_preprocessor.scope_name] = self._pi_v_state_preprocessor
return models
def _solvers(self):
solvers = {}
solvers[self._v_function.scope_name] = self._v_function_solver
solvers[self._discriminator.scope_name] = self._discriminator_solver
return solvers
[docs] @classmethod
def is_supported_env(cls, env_or_env_info):
env_info = EnvironmentInfo.from_env(env_or_env_info) if isinstance(env_or_env_info, gym.Env) \
else env_or_env_info
return not env_info.is_discrete_action_env() and not env_info.is_tuple_action_env()
@property
def latest_iteration_state(self):
latest_iteration_state = super(GAIL, self).latest_iteration_state
if hasattr(self, '_discriminator_trainer_state'):
latest_iteration_state['scalar'].update(
{'reward_loss': float(self._discriminator_trainer_state['reward_loss'])})
if hasattr(self, '_v_function_trainer_state'):
latest_iteration_state['scalar'].update({'v_loss': float(self._v_function_trainer_state['v_loss'])})
return latest_iteration_state
@property
def trainers(self):
return {
"discriminator": self._discriminator_trainer,
"v_function": self._v_function_trainer,
"policy": self._policy_trainer,
}