f80d8818f159be225c622d29ce705e6f52839f6d,tensorforce/models/random_model.py,RandomModel,tf_actions_and_internals,#RandomModel#Any#Any#Any#,29

Before Change

                        maxval=action.max_value,
                    )
                else:
                    actions[name] = tf.random_normal(shape=action.shape, name="random-action")

        return actions, internals

    def tf_loss_per_instance(self, states, internals, actions, terminal, reward):

After Change

            print(states.values())
            print(type(next(iter(states.values()))))
            print(tf.shape(input=next(iter(states.values()))))
            shape = tuple(tf.shape(input=next(iter(states.values())))[0],) + action["shape"]

            if action["type"] == "bool":
                action = tf.random_uniform(shape=shape)

Instances