saymrwulf/stable-baselines3
1
0
Fork 0
mirror of https://github.com/saymrwulf/stable-baselines3.git synced 2026-07-30 20:18:15 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Bug fix actor forward

Browse source
This commit is contained in:
Antonin RAFFIN 2019-09-18 23:55:41 +02:00
parent 525fe43552
commit 149148d4c7
2 changed files with 8 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

10
torchy_baselines/ppo/policies.py
View file

@ -55,13 +55,17 @@ class PPOPolicy(BasePolicy):
value = self.value_net(latent)
return action, value, log_prob
def actor_forward(self, state):
def actor_forward(self, state, deterministic=False):
latent = self.shared_net(state)
# TODO: initialize pi_mean weights properly
mean_actions = self.actor_net(latent)
action_distribution = Normal(mean_actions, self.log_std)
action_std = th.ones(mean_actions.size()) * self.log_std.exp()
action_distribution = Normal(mean_actions, action_std)
# Sample from the gaussian
action = action_distribution.rsample()
if deterministic:
action = mean_actions
else:
action = action_distribution.rsample()
return action
def get_policy_stats(self, state, action):

2
torchy_baselines/ppo/ppo.py
View file

@ -58,7 +58,7 @@ class PPO(BaseRLModel):
observation = np.array(observation)
with th.no_grad():
observation = th.FloatTensor(observation.reshape(1, -1)).to(self.device)
return self.policy.actor_forward(observation).cpu().data.numpy().flatten()
return self.policy.actor_forward(observation, deterministic=False).cpu().data.numpy().flatten()
def predict(self, observation, state=None, mask=None, deterministic=True):
"""