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

modified training step display

Browse Source
This commit is contained in:
NM512 2023-06-24 23:05:45 +09:00
parent f3fe3a872e
commit 34a44916f7
2 changed files with 22 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
dreamer.py
View File

@ -213,10 +213,12 @@ def make_env(config, logger, mode, train_eps, eval_eps):
env = wrappers.OneHotAction(env) env = wrappers.OneHotAction(env)
elif suite == "MemoryMaze": elif suite == "MemoryMaze":
from envs.memorymaze import MemoryMaze from envs.memorymaze import MemoryMaze
env = MemoryMaze(task) env = MemoryMaze(task)
env = wrappers.OneHotAction(env) env = wrappers.OneHotAction(env)
elif suite == "crafter": elif suite == "crafter":
import envs.crafter as crafter import envs.crafter as crafter
env = crafter.Crafter(task, config.size) env = crafter.Crafter(task, config.size)
env = wrappers.OneHotAction(env) env = wrappers.OneHotAction(env)
else: else:
@ -254,17 +256,19 @@ class ProcessEpisodeWrap:
length = len(episode["reward"]) - 1 length = len(episode["reward"]) - 1
score = float(episode["reward"].astype(np.float64).sum()) score = float(episode["reward"].astype(np.float64).sum())
video = episode["image"] video = episode["image"]
# add new episode
cache[str(filename)] = episode cache[str(filename)] = episode
if mode == "train": if mode == "train":
total = 0 step_in_dataset = 0
for key, ep in reversed(sorted(cache.items(), key=lambda x: x[0])): for key, ep in reversed(sorted(cache.items(), key=lambda x: x[0])):
if not config.dataset_size or total <= config.dataset_size - length: if (
total += len(ep["reward"]) - 1 not config.dataset_size
or step_in_dataset + (len(ep["reward"]) - 1) <= config.dataset_size
):
step_in_dataset += len(ep["reward"]) - 1
else: else:
del cache[key] del cache[key]
logger.scalar("dataset_size", total) logger.scalar("dataset_size", step_in_dataset)
# use dataset_size as log step for a condition of envs > 1
log_step = total * config.action_repeat
elif mode == "eval": elif mode == "eval":
# keep only last item for saving memory # keep only last item for saving memory
while len(cache) > 1: while len(cache) > 1:
@ -285,7 +289,6 @@ class ProcessEpisodeWrap:
score = sum(cls.eval_scores) / len(cls.eval_scores) score = sum(cls.eval_scores) / len(cls.eval_scores)
length = sum(cls.eval_lengths) / len(cls.eval_lengths) length = sum(cls.eval_lengths) / len(cls.eval_lengths)
episode_num = len(cls.eval_scores) episode_num = len(cls.eval_scores)
log_step = logger.step
logger.video(f"{mode}_policy", video[None]) logger.video(f"{mode}_policy", video[None])
cls.eval_done = True cls.eval_done = True
@ -295,7 +298,7 @@ class ProcessEpisodeWrap:
logger.scalar( logger.scalar(
f"{mode}_episodes", len(cache) if mode == "train" else episode_num f"{mode}_episodes", len(cache) if mode == "train" else episode_num
) )
logger.write(step=log_step) logger.write(step=logger.step)
def main(config): def main(config):

9
envs/memorymaze.py
View File

@ -32,13 +32,16 @@ class MemoryMaze:
spaces = self._env.observation_space.spaces.copy() spaces = self._env.observation_space.spaces.copy()
else: else:
spaces = {self._obs_key: self._env.observation_space} spaces = {self._obs_key: self._env.observation_space}
return gym.spaces.Dict({ return gym.spaces.Dict(
{
**spaces, **spaces,
"reward": gym.spaces.Box(-np.inf, np.inf, (), dtype=np.float32), "reward": gym.spaces.Box(-np.inf, np.inf, (), dtype=np.float32),
"is_first": gym.spaces.Box(0, 1, (), dtype=bool), "is_first": gym.spaces.Box(0, 1, (), dtype=bool),
"is_last": gym.spaces.Box(0, 1, (), dtype=bool), "is_last": gym.spaces.Box(0, 1, (), dtype=bool),
"is_terminal": gym.spaces.Box(0, 1, (), dtype=bool), "is_terminal": gym.spaces.Box(0, 1, (), dtype=bool),
}) }
)
@property @property
def action_space(self): def action_space(self):
space = self._env.action_space space = self._env.action_space
@ -49,7 +52,7 @@ class MemoryMaze:
obs, reward, done, info = self._env.step(action) obs, reward, done, info = self._env.step(action)
if not self._obs_is_dict: if not self._obs_is_dict:
obs = {self._obs_key: obs} obs = {self._obs_key: obs}
obs['reward'] = reward obs["reward"] = reward
obs["is_first"] = False obs["is_first"] = False
obs["is_last"] = done obs["is_last"] = done
obs["is_terminal"] = info.get("is_terminal", False) obs["is_terminal"] = info.get("is_terminal", False)