f353943f836ad162ae45a96a404d5db48dba7690,ml/rl/training/parametric_dqn_trainer.py,ParametricDQNTrainer,internal_prediction,#ParametricDQNTrainer#Any#Any#,137

Before Change

        
        self.q_network.eval()
        q_values = self.q_network(
            rlt.StateAction(
                state=rlt.FeatureVector(float_features=state),
                action=rlt.FeatureVector(float_features=action),
            )
        )
        self.q_network.train()
        return q_values.q_value.cpu()

After Change

        
        self.q_network.eval()
        q_values = self.q_network(
            rlt.PreprocessedStateAction(state=state, action=action)
        )
        self.q_network.train()
        return q_values.q_value.cpu()

Instances