Spaces:
Sleeping
Sleeping
File size: 6,132 Bytes
eaf2e33 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 |
from gym.spaces import Discrete
from src.rlkit.data_management.simple_replay_buffer import SimpleReplayBuffer, EnsembleSimpleReplayBuffer
from src.rlkit.data_management.simple_replay_buffer import RandomReplayBuffer, GaussianReplayBuffer
from src.rlkit.envs.env_utils import get_dim
import numpy as np
class EnvReplayBuffer(SimpleReplayBuffer):
def __init__(
self,
max_replay_buffer_size,
env,
env_info_sizes=None
):
"""
:param max_replay_buffer_size:
:param env:
"""
self.env = env
self._ob_space = env.observation_space
self._action_space = env.action_space
if env_info_sizes is None:
if hasattr(env, 'info_sizes'):
env_info_sizes = env.info_sizes
else:
env_info_sizes = dict()
super().__init__(
max_replay_buffer_size=max_replay_buffer_size,
observation_dim=get_dim(self._ob_space),
action_dim=get_dim(self._action_space),
env_info_sizes=env_info_sizes
)
def add_sample(self, observation, action, reward, terminal,
next_observation, **kwargs):
if isinstance(self._action_space, Discrete):
new_action = np.zeros(self._action_dim)
new_action[action] = 1
else:
new_action = action
return super().add_sample(
observation=observation,
action=new_action,
reward=reward,
next_observation=next_observation,
terminal=terminal,
**kwargs
)
class EnsembleEnvReplayBuffer(EnsembleSimpleReplayBuffer):
def __init__(
self,
max_replay_buffer_size,
env,
num_ensemble,
log_dir,
env_info_sizes=None
):
"""
:param max_replay_buffer_size:
:param env:
"""
self.env = env
self._ob_space = env.observation_space
self._action_space = env.action_space
if env_info_sizes is None:
if hasattr(env, 'info_sizes'):
env_info_sizes = env.info_sizes
else:
env_info_sizes = dict()
super().__init__(
max_replay_buffer_size=max_replay_buffer_size,
observation_dim=get_dim(self._ob_space),
action_dim=get_dim(self._action_space),
env_info_sizes=env_info_sizes,
num_ensemble=num_ensemble,
log_dir=log_dir,
)
def add_sample(self, observation, action, reward, terminal,
next_observation, mask, **kwargs):
if isinstance(self._action_space, Discrete):
new_action = np.zeros(self._action_dim)
new_action[action] = 1
else:
new_action = action
return super().add_sample(
observation=observation,
action=new_action,
reward=reward,
next_observation=next_observation,
terminal=terminal,
mask=mask,
**kwargs
)
class RandomEnvReplayBuffer(RandomReplayBuffer):
def __init__(
self,
max_replay_buffer_size,
env,
single_flag,
equal_flag,
lower,
upper,
env_info_sizes=None
):
"""
:param max_replay_buffer_size:
:param env:
"""
self.env = env
self._ob_space = env.observation_space
self._action_space = env.action_space
if env_info_sizes is None:
if hasattr(env, 'info_sizes'):
env_info_sizes = env.info_sizes
else:
env_info_sizes = dict()
super().__init__(
max_replay_buffer_size=max_replay_buffer_size,
observation_dim=get_dim(self._ob_space),
action_dim=get_dim(self._action_space),
env_info_sizes=env_info_sizes,
single_flag=single_flag,
equal_flag=equal_flag,
lower=lower,
upper=upper,
)
def add_sample(self, observation, action, reward, terminal,
next_observation, **kwargs):
if isinstance(self._action_space, Discrete):
new_action = np.zeros(self._action_dim)
new_action[action] = 1
else:
new_action = action
return super().add_sample(
observation=observation,
action=new_action,
reward=reward,
next_observation=next_observation,
terminal=terminal,
**kwargs
)
class GaussianEnvReplayBuffer(GaussianReplayBuffer):
def __init__(
self,
max_replay_buffer_size,
env,
prob,
std,
env_info_sizes=None
):
"""
:param max_replay_buffer_size:
:param env:
"""
self.env = env
self._ob_space = env.observation_space
self._action_space = env.action_space
if env_info_sizes is None:
if hasattr(env, 'info_sizes'):
env_info_sizes = env.info_sizes
else:
env_info_sizes = dict()
super().__init__(
max_replay_buffer_size=max_replay_buffer_size,
observation_dim=get_dim(self._ob_space),
action_dim=get_dim(self._action_space),
env_info_sizes=env_info_sizes,
prob=prob,
std=std,
)
def add_sample(self, observation, action, reward, terminal,
next_observation, **kwargs):
if isinstance(self._action_space, Discrete):
new_action = np.zeros(self._action_dim)
new_action[action] = 1
else:
new_action = action
return super().add_sample(
observation=observation,
action=new_action,
reward=reward,
next_observation=next_observation,
terminal=terminal,
**kwargs
)
|