e72838c03d87c8eb10f7cb8df8804496b14121ab,rllib/agents/marwil/marwil_torch_policy.py,,setup_mixins,#,70

Before Change

def setup_mixins(policy, obs_space, action_space, config):
    // Create a var.
    policy.ma_adv_norm = torch.tensor(
        [100.0], dtype=torch.float32, requires_grad=False)
    // Setup Value branch of our NN.
    ValueNetworkMixin.__init__(policy)

After Change

def setup_mixins(policy, obs_space, action_space, config):
    // Create a var.
    policy.ma_adv_norm = torch.tensor(
        [100.0], dtype=torch.float32, requires_grad=False).to(policy.device)
    // Setup Value branch of our NN.
    ValueNetworkMixin.__init__(policy)

Instances