OscervaME
/
stable-baselines3-contrib-sacd
2
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Use higher resolution time_ns() and avoid division by zero (#91) 

* Use higher resolution time_ns and add max to avoid division by zero

* Add missing imports

* Update changelog
This commit is contained in:
Adam Gleave 2022-07-25 14:12:20 -07:00 committed by GitHub
parent 3cbd2429be
commit 7e687ac47c
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
3 changed files with 9 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

1
docs/misc/changelog.rst
View File

@ -15,6 +15,7 @@ New Features:
Bug Fixes: Bug Fixes:
^^^^^^^^^^ ^^^^^^^^^^
- Fixed division by zero error when computing FPS when a small number of time has elapsed in operating systems with low-precision timers.
Deprecations: Deprecations:
^^^^^^^^^^^^^ ^^^^^^^^^^^^^

5
sb3_contrib/ars/ars.py
View File

@ -1,4 +1,5 @@
import copy import copy
import sys
import time import time
import warnings import warnings
from functools import partial from functools import partial
@ -242,8 +243,8 @@ class ARS(BaseAlgorithm):
""" """
Dump information to the logger. Dump information to the logger.
""" """
time_elapsed = time.time() - self.start_time time_elapsed = max((time.time_ns() - self.start_time) / 1e9, sys.float_info.epsilon)
fps = int((self.num_timesteps - self._num_timesteps_at_start) / (time_elapsed + 1e-8)) fps = int((self.num_timesteps - self._num_timesteps_at_start) / time_elapsed)
if len(self.ep_info_buffer) > 0 and len(self.ep_info_buffer[0]) > 0: if len(self.ep_info_buffer) > 0 and len(self.ep_info_buffer[0]) > 0:
self.logger.record("rollout/ep_rew_mean", safe_mean([ep_info["r"] for ep_info in self.ep_info_buffer])) self.logger.record("rollout/ep_rew_mean", safe_mean([ep_info["r"] for ep_info in self.ep_info_buffer]))
self.logger.record("rollout/ep_len_mean", safe_mean([ep_info["l"] for ep_info in self.ep_info_buffer])) self.logger.record("rollout/ep_len_mean", safe_mean([ep_info["l"] for ep_info in self.ep_info_buffer]))

8
sb3_contrib/ppo_mask/ppo_mask.py
View File

@ -1,3 +1,4 @@
import sys
import time import time
from collections import deque from collections import deque
from typing import Any, Dict, Optional, Tuple, Type, Union from typing import Any, Dict, Optional, Tuple, Type, Union
@ -242,7 +243,7 @@ class MaskablePPO(OnPolicyAlgorithm):
:return: :return:
""" """
self.start_time = time.time() self.start_time = time.time_ns()
if self.ep_info_buffer is None or reset_num_timesteps: if self.ep_info_buffer is None or reset_num_timesteps:
# Initialize buffers if they don't exist, or reinitialize if resetting counters # Initialize buffers if they don't exist, or reinitialize if resetting counters
self.ep_info_buffer = deque(maxlen=100) self.ep_info_buffer = deque(maxlen=100)
@ -566,13 +567,14 @@ class MaskablePPO(OnPolicyAlgorithm):
# Display training infos # Display training infos
if log_interval is not None and iteration % log_interval == 0: if log_interval is not None and iteration % log_interval == 0:
fps = int((self.num_timesteps - self._num_timesteps_at_start) / (time.time() - self.start_time)) time_elapsed = max((time.time_ns() - self.start_time) / 1e9, sys.float_info.epsilon)
fps = int((self.num_timesteps - self._num_timesteps_at_start) / time_elapsed)
self.logger.record("time/iterations", iteration, exclude="tensorboard") self.logger.record("time/iterations", iteration, exclude="tensorboard")
if len(self.ep_info_buffer) > 0 and len(self.ep_info_buffer[0]) > 0: if len(self.ep_info_buffer) > 0 and len(self.ep_info_buffer[0]) > 0:
self.logger.record("rollout/ep_rew_mean", safe_mean([ep_info["r"] for ep_info in self.ep_info_buffer])) self.logger.record("rollout/ep_rew_mean", safe_mean([ep_info["r"] for ep_info in self.ep_info_buffer]))
self.logger.record("rollout/ep_len_mean", safe_mean([ep_info["l"] for ep_info in self.ep_info_buffer])) self.logger.record("rollout/ep_len_mean", safe_mean([ep_info["l"] for ep_info in self.ep_info_buffer]))
self.logger.record("time/fps", fps) self.logger.record("time/fps", fps)
self.logger.record("time/time_elapsed", int(time.time() - self.start_time), exclude="tensorboard") self.logger.record("time/time_elapsed", int(time_elapsed), exclude="tensorboard")
self.logger.record("time/total_timesteps", self.num_timesteps, exclude="tensorboard") self.logger.record("time/total_timesteps", self.num_timesteps, exclude="tensorboard")
self.logger.dump(step=self.num_timesteps) self.logger.dump(step=self.num_timesteps)