Spaces:
Sleeping
Sleeping
File size: 6,522 Bytes
375a1cf |
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 |
"""Contains methods for step compatibility, from old-to-new and new-to-old API."""
from typing import Tuple, Union
import numpy as np
from gym.core import ObsType
DoneStepType = Tuple[
Union[ObsType, np.ndarray],
Union[float, np.ndarray],
Union[bool, np.ndarray],
Union[dict, list],
]
TerminatedTruncatedStepType = Tuple[
Union[ObsType, np.ndarray],
Union[float, np.ndarray],
Union[bool, np.ndarray],
Union[bool, np.ndarray],
Union[dict, list],
]
def convert_to_terminated_truncated_step_api(
step_returns: Union[DoneStepType, TerminatedTruncatedStepType], is_vector_env=False
) -> TerminatedTruncatedStepType:
"""Function to transform step returns to new step API irrespective of input API.
Args:
step_returns (tuple): Items returned by step(). Can be (obs, rew, done, info) or (obs, rew, terminated, truncated, info)
is_vector_env (bool): Whether the step_returns are from a vector environment
"""
if len(step_returns) == 5:
return step_returns
else:
assert len(step_returns) == 4
observations, rewards, dones, infos = step_returns
# Cases to handle - info single env / info vector env (list) / info vector env (dict)
if is_vector_env is False:
truncated = infos.pop("TimeLimit.truncated", False)
return (
observations,
rewards,
dones and not truncated,
dones and truncated,
infos,
)
elif isinstance(infos, list):
truncated = np.array(
[info.pop("TimeLimit.truncated", False) for info in infos]
)
return (
observations,
rewards,
np.logical_and(dones, np.logical_not(truncated)),
np.logical_and(dones, truncated),
infos,
)
elif isinstance(infos, dict):
num_envs = len(dones)
truncated = infos.pop("TimeLimit.truncated", np.zeros(num_envs, dtype=bool))
return (
observations,
rewards,
np.logical_and(dones, np.logical_not(truncated)),
np.logical_and(dones, truncated),
infos,
)
else:
raise TypeError(
f"Unexpected value of infos, as is_vector_envs=False, expects `info` to be a list or dict, actual type: {type(infos)}"
)
def convert_to_done_step_api(
step_returns: Union[TerminatedTruncatedStepType, DoneStepType],
is_vector_env: bool = False,
) -> DoneStepType:
"""Function to transform step returns to old step API irrespective of input API.
Args:
step_returns (tuple): Items returned by step(). Can be (obs, rew, done, info) or (obs, rew, terminated, truncated, info)
is_vector_env (bool): Whether the step_returns are from a vector environment
"""
if len(step_returns) == 4:
return step_returns
else:
assert len(step_returns) == 5
observations, rewards, terminated, truncated, infos = step_returns
# Cases to handle - info single env / info vector env (list) / info vector env (dict)
if is_vector_env is False:
if truncated or terminated:
infos["TimeLimit.truncated"] = truncated and not terminated
return (
observations,
rewards,
terminated or truncated,
infos,
)
elif isinstance(infos, list):
for info, env_truncated, env_terminated in zip(
infos, truncated, terminated
):
if env_truncated or env_terminated:
info["TimeLimit.truncated"] = env_truncated and not env_terminated
return (
observations,
rewards,
np.logical_or(terminated, truncated),
infos,
)
elif isinstance(infos, dict):
if np.logical_or(np.any(truncated), np.any(terminated)):
infos["TimeLimit.truncated"] = np.logical_and(
truncated, np.logical_not(terminated)
)
return (
observations,
rewards,
np.logical_or(terminated, truncated),
infos,
)
else:
raise TypeError(
f"Unexpected value of infos, as is_vector_envs=False, expects `info` to be a list or dict, actual type: {type(infos)}"
)
def step_api_compatibility(
step_returns: Union[TerminatedTruncatedStepType, DoneStepType],
output_truncation_bool: bool = True,
is_vector_env: bool = False,
) -> Union[TerminatedTruncatedStepType, DoneStepType]:
"""Function to transform step returns to the API specified by `output_truncation_bool` bool.
Done (old) step API refers to step() method returning (observation, reward, done, info)
Terminated Truncated (new) step API refers to step() method returning (observation, reward, terminated, truncated, info)
(Refer to docs for details on the API change)
Args:
step_returns (tuple): Items returned by step(). Can be (obs, rew, done, info) or (obs, rew, terminated, truncated, info)
output_truncation_bool (bool): Whether the output should return two booleans (new API) or one (old) (True by default)
is_vector_env (bool): Whether the step_returns are from a vector environment
Returns:
step_returns (tuple): Depending on `output_truncation_bool` bool, it can return (obs, rew, done, info) or (obs, rew, terminated, truncated, info)
Examples:
This function can be used to ensure compatibility in step interfaces with conflicting API. Eg. if env is written in old API,
wrapper is written in new API, and the final step output is desired to be in old API.
>>> obs, rew, done, info = step_api_compatibility(env.step(action), output_truncation_bool=False)
>>> obs, rew, terminated, truncated, info = step_api_compatibility(env.step(action), output_truncation_bool=True)
>>> observations, rewards, dones, infos = step_api_compatibility(vec_env.step(action), is_vector_env=True)
"""
if output_truncation_bool:
return convert_to_terminated_truncated_step_api(step_returns, is_vector_env)
else:
return convert_to_done_step_api(step_returns, is_vector_env)
|