59565bb8a0759cf2fc2c264c59e1ee29daff5fe3,models/mlp_critic.py,Value,forward,#Value#Any#,15

Before Change

        self.value_head.bias.data.mul_(0.0)

    def forward(self, x):
        x = F.tanh(self.affine1(x))
        x = F.tanh(self.affine2(x))

        value = self.value_head(x)
        return value

After Change

        self.value_head.weight.data.mul_(0.1)
        self.value_head.bias.data.mul_(0.0)

    def forward(self, x):
        for affine in self.affine_layers:
            x = self.activation(affine(x))

        value = self.value_head(x)
        return value

Instances