d1ae17e7f6a0916add0d2b4521acab3fbb42650e,tensorforce/core/models/model.py,Model,tf_initialize,#Model#,615

Before Change

                    ), axis=0
                )
                default = tf.ones(shape=shape, dtype=util.tf_dtype(dtype="bool"))
                shape = action_spec["shape"] + (action_spec["num_values"],)
                self.auxiliaries_input[name] = self.add_placeholder(
                    name=name, dtype="bool", shape=shape, batched=True, default=default
                )

After Change

                    )[:1]
                else:
                    batch_size = tf.dtypes.cast(
                        x=tf.shape(input=next(iter(self.states_input.values())))[:1],
                        dtype=util.tf_dtype(dtype="long")
                    )
                tf_shape = tf.constant(
                    value=(action_spec["shape"] + (action_spec["num_values"],)),
                    dtype=util.tf_dtype(dtype="long")
                )
                tf_shape = tf.concat(values=(batch_size, tf_shape), axis=0)
                default = tf.ones(shape=((1,) + shape), dtype=util.tf_dtype(dtype="bool"))
                self.auxiliaries_input[name] = self.add_placeholder(
                    name=name, dtype="bool", shape=shape, batched=True, default=default
                )

Instances