db09dc1fb503ab8f7de69fa23e8d38742bda8e90,ch07/03_dqn_double.py,,calc_values_of_states,#Any#Any#Any#,48

Before Change

def calc_values_of_states(states, net, cuda=False):
    mean_vals = []
    for batch in np.array_split(states, 64):
        states_v = Variable(torch.from_numpy(batch), volatile=True)
        if cuda:
            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

def calc_values_of_states(states, net, device="cpu"):
    mean_vals = []
    for batch in np.array_split(states, 64):
        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