wassname/pytorch-soft-actor-critic
1
0
Fork 0
mirror of https://github.com/wassname/pytorch-soft-actor-critic.git synced 2026-08-12 12:20:51 +08:00
Code Issues Packages Projects Releases Wiki Activity

Update model.py

Browse Source
This commit is contained in:
Pranjal Tandon
2018-09-01 19:03:04 +05:30
committed by GitHub
parent a3f9f3821f
commit 086f9fe36c
1 changed files with 4 additions and 15 deletions
Download Patch File Download Diff File Expand all files Collapse all files
model.py
+4 -15
View File
@@ -11,16 +11,13 @@ LOG_SIG_MIN = -20
class ValueNetwork(nn.Module):
def __init__(self, state_dim, hidden_dim, init_w=1e-3):
def __init__(self, state_dim, hidden_dim):
super(ValueNetwork, self).__init__()
self.linear1 = nn.Linear(state_dim, hidden_dim)
self.linear2 = nn.Linear(hidden_dim, hidden_dim)
self.linear3 = nn.Linear(hidden_dim, 1)
self.linear3.weight.data.uniform_(-init_w, init_w)
self.linear3.bias.data.uniform_(-init_w, init_w)
def forward(self, state):
x = F.relu(self.linear1(state))
x = F.relu(self.linear2(x))
@@ -29,16 +26,13 @@ class ValueNetwork(nn.Module):
class QNetwork(nn.Module):
def __init__(self, num_inputs, num_actions, hidden_size, init_w=1e-3):
def __init__(self, num_inputs, num_actions, hidden_size):
super(QNetwork, self).__init__()
self.linear1 = nn.Linear(num_inputs + num_actions, hidden_size)
self.linear2 = nn.Linear(hidden_size, hidden_size)
self.linear3 = nn.Linear(hidden_size, 1)
self.linear3.weight.data.uniform_(-init_w, init_w)
self.linear3.bias.data.uniform_(-init_w, init_w)
def forward(self, state, action):
x = torch.cat([state, action], 1)
x = F.relu(self.linear1(x))
@@ -48,19 +42,15 @@ class QNetwork(nn.Module):
class GaussianPolicy(nn.Module):
def __init__(self, num_inputs, num_actions, hidden_size, init_w=1e-3):
def __init__(self, num_inputs, num_actions, hidden_size):
super(GaussianPolicy, self).__init__()
self.linear1 = nn.Linear(num_inputs, hidden_size)
self.linear2 = nn.Linear(hidden_size, hidden_size)
self.mean_linear = nn.Linear(hidden_size, num_actions)
self.mean_linear.weight.data.uniform_(-init_w, init_w)
self.mean_linear.bias.data.uniform_(-init_w, init_w)
self.log_std_linear = nn.Linear(hidden_size, num_actions)
self.log_std_linear.weight.data.uniform_(-init_w, init_w)
self.log_std_linear.bias.data.uniform_(-init_w, init_w)
def forward(self, state):
x = F.relu(self.linear1(state))
@@ -79,7 +69,7 @@ class GaussianPolicy(nn.Module):
normal = Normal(mean, std)
if reparam == True:
x_t = normal.rsample() # or mean + std * torch.randn(1,6)
x_t = normal.rsample() #mean + std * torch.randn(1,6)
else:
x_t = normal.sample()
@@ -103,7 +93,6 @@ class GaussianMixturePolicy(nn.Module):
super(GaussianMixturePolicy, self).__init__()
self.actions = num_actions
self.k = k
self.log_std_max = LOG_SIG_MAX
self.linear1 = nn.Linear(num_inputs, hidden_size)
self.linear2 = nn.Linear(hidden_size, hidden_size)