saymrwulf/stable-baselines3
1
0
Fork 0
mirror of https://github.com/saymrwulf/stable-baselines3.git synced 2026-07-10 17:37:31 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Fix deterministic run

Browse source
This commit is contained in:
Antonin RAFFIN 2019-09-21 15:53:28 +02:00
parent a9b8276efb
commit bcdd99d22c
1 changed files with 5 additions and 2 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

7
torchy_baselines/common/base_class.py
View file

@ -210,8 +210,12 @@ class BaseRLModel(object):
def seed(self, seed=0):
set_random_seed(seed, using_cuda=self.device == th.device('cuda'))
self.action_space.seed(seed)
if self.env is not None:
self.env.seed(seed)
if self.eval_env is not None:
self.eval_env.seed(seed)
def collect_rollouts(self, env, n_episodes=1, action_noise_std=0.0,
deterministic=False, callback=None, remove_timelimits=True,
@ -219,7 +223,6 @@ class BaseRLModel(object):
episode_rewards = []
total_timesteps = []
assert isinstance(env, VecEnv)
assert env.num_envs == 1
@ -231,7 +234,7 @@ class BaseRLModel(object):
while not done:
# Select action randomly or according to policy
if num_timesteps < start_timesteps:
action = [env.action_space.sample()]
action = [self.action_space.sample()]
else:
action = self.predict(obs, deterministic=deterministic) / self.max_action