hongshaorou/dreamerv3-torch
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

added metadata and reward_range for each tasks

Browse Source
This commit is contained in:
NM512 2023-07-23 21:55:22 +09:00
parent 106317015d
commit 2c933da684
3 changed files with 6 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
envs/atari.py
View File

@ -4,7 +4,7 @@ import numpy as np
class Atari: class Atari:
LOCK = None LOCK = None
metadata = {}
def __init__( def __init__(
self, self,
name, name,
@ -64,6 +64,7 @@ class Atari:
self._last_lives = None self._last_lives = None
self._done = True self._done = True
self._step = 0 self._step = 0
self.reward_range = [-np.inf, np.inf]
@property @property
def observation_space(self): def observation_space(self):

2
envs/crafter.py
View File

@ -3,12 +3,14 @@ import numpy as np
class Crafter: class Crafter:
metadata = {}
def __init__(self, task, size=(64, 64), seed=None): def __init__(self, task, size=(64, 64), seed=None):
assert task in ("reward", "noreward") assert task in ("reward", "noreward")
import crafter import crafter
self._env = crafter.Env(size=size, reward=(task == "reward"), seed=seed) self._env = crafter.Env(size=size, reward=(task == "reward"), seed=seed)
self._achievements = crafter.constants.achievements.copy() self._achievements = crafter.constants.achievements.copy()
self.reward_range = [-np.inf, np.inf]
@property @property
def observation_space(self): def observation_space(self):

2
envs/dmc.py
View File

@ -3,6 +3,7 @@ import numpy as np
class DeepMindControl: class DeepMindControl:
metadata = {}
def __init__(self, name, action_repeat=1, size=(64, 64), camera=None): def __init__(self, name, action_repeat=1, size=(64, 64), camera=None):
domain, task = name.split("_", 1) domain, task = name.split("_", 1)
if domain == "cup": # Only domain with multiple words. if domain == "cup": # Only domain with multiple words.
@ -19,6 +20,7 @@ class DeepMindControl:
if camera is None: if camera is None:
camera = dict(quadruped=2).get(domain, 0) camera = dict(quadruped=2).get(domain, 0)
self._camera = camera self._camera = camera
self.reward_range = [-np.inf, np.inf]
@property @property
def observation_space(self): def observation_space(self):