8dcbf002d0058939802920c11f74791708838791,ml/rl/training/ddpg_predictor.py,DDPGPredictor,predict_q_value,#DDPGPredictor#Any#Any#,31

Before Change

            for k, v in actions[i].items():
                action[k - len(state)] = v
            examples.append((state, action))
        output = [self.critic(example[0], example[1]) for example in examples]
        return output

    @classmethod
    def export_actor(cls, trainer):
        return DDPGPredictor(trainer)

After Change

            for k, v in actions[i].items():
                action[k - len(state)] = v
            action_list.append(action)
        output = self.critic(
            Variable(torch.from_numpy(np.array(state_list))),
            Variable(torch.from_numpy(np.array(action_list)))
        )
        return output.data.numpy()

    @classmethod
    def export_actor(cls, trainer):
        return DDPGPredictor(trainer)

Instances