b5e848af46b4a6fc21e718803dbf3d7c72afd72a,ch08/lib/common.py,,calc_values_of_states,#,63

Before Change

            states_v = states_v.cuda()
        action_values_v = net(states_v)
        best_action_values_v = action_values_v.max(1)[0]
        mean_val = best_action_values_v.mean().data.cpu().numpy()[0]
        mean_vals.append(mean_val)
    return np.mean(mean_vals)

After Change

        states_v = torch.tensor(batch).to(device)
        action_values_v = net(states_v)
        best_action_values_v = action_values_v.max(1)[0]
        mean_vals.append(best_action_values_v.mean().item())
    return np.mean(mean_vals)

Instances