wassname/pytorch-soft-actor-critic
1
0
Fork 0
mirror of https://github.com/wassname/pytorch-soft-actor-critic.git synced 2026-08-15 12:45:26 +08:00
Code Issues Packages Projects Releases Wiki Activity

Update sac.py

Browse Source
This commit is contained in:
Pranjal Tandon
2018-10-04 11:23:30 +05:30
committed by GitHub
parent 41577fa4d4
commit 3b60dd065c
1 changed files with 11 additions and 7 deletions
Download Patch File Download Diff File Expand all files Collapse all files
sac.py
+11 -7
View File
@@ -38,15 +38,19 @@ class SAC(object):
self.soft_q_criterion = nn.MSELoss()
def select_action(self, state, deterministic=False):
def select_action(self, state, eval=False):
state = torch.FloatTensor(state).unsqueeze(0)
if deterministic == False:
_, _, x_t, _, _ = self.policy.evaluate(state)
action = torch.tanh(x_t)
if eval == False:
self.policy.train()
print("train")
_, _, action, _, _ = self.policy.evaluate(state)
else:
_, _, _, x_t, _ = self.policy.evaluate(state)
action = torch.tanh(x_t)
action = action.detach().numpy()
self.policy.eval()
print("eval")
_, _, _, action, _ = self.policy.evaluate(state)
action = torch.tanh(action)
action = action.detach().cpu().numpy()
return action[0]