def __init__(self, state_size):
super(Value, self).__init__()
self.state_size = state_size
self.fc1 = rn.NoisyLinear(state_size, 64)
self.fc_norm = nn.LayerNorm(64)
self.fc2 = rn.NoisyLinear(64, 64)
self.fc2_norm = nn.LayerNorm(64)
self.fc3 = rn.NoisyLinear(64, 1)
def __init__(self, state_size, action_size):
super(Policy, self).__init__()
self.state_size = state_size
self.action_size = action_size
self.fc1 = rn.NoisyLinear(state_size, 64)
self.fc_norm = nn.LayerNorm(64)
self.fc2 = rn.NoisyLinear(64, 64)
self.fc2_norm = nn.LayerNorm(64)
self.fc3 = rn.NoisyLinear(64, action_size)
def __init__(self, state_size, action_size):
super(Value, self).__init__()
self.state_size = state_size
self.action_size = action_size
self.fc1 = rn.NoisyLinear(state_size, 255)
self.fc_norm = nn.LayerNorm(255)
self.value_fc = rn.NoisyLinear(255, 255)
self.value_fc_norm = nn.LayerNorm(255)
self.value = rn.NoisyLinear(255, 1)
self.advantage_fc = rn.NoisyLinear(255, 255)
self.advantage_fc_norm = nn.LayerNorm(255)
self.advantage = rn.NoisyLinear(255, action_size)
def __init__(self, state_size, action_size):
super(Value, self).__init__()
self.state_size = state_size
self.action_size = action_size
self.conv1 = nn.Conv2d(4, 32, kernel_size=(8, 8), stride=(4, 4))
self.conv2 = nn.Conv2d(32, 64, kernel_size=(4, 4), stride=(2, 2))
self.conv3 = nn.Conv2d(64, 64, kernel_size=(3, 3), stride=(1, 1))
self.fc1 = nn.Linear(3136, 512)
self.fc1_norm = nn.LayerNorm(512)
self.value_fc = rn.NoisyLinear(512, 512)
self.value_fc_norm = nn.LayerNorm(512)
self.value = nn.Linear(512, 1)
self.advantage_fc = rn.NoisyLinear(512, 512)
self.advantage_fc_norm = nn.LayerNorm(512)
self.advantage = nn.Linear(512, action_size)