bcf723a2f67a372e6b03a453149a82ecf2766fff,theanolm/network.py,Architecture,check_state,#Architecture#Any#,118

Before Change

                    "Parameter "arch.layers" is missing from neural network state.")
            // An ugly workaround to be able to save arbitrary data in a .npz file.
            try:
                dummy_dict = state["arch.layers"][()]
            except KeyError:
                dummy_dict = state["arch.layers"]
            state_layers = dummy_dict["data"]
            for layer1, layer2 in zip(self.layers, state_layers):
                if layer1["type"] != layer2["type"]:
                    raise IncompatibleStateError(
                        "Neural network state has {0}={1}, while this architecture "
                        "has {0}={2}.".format("type", layer2["type"], layer1["type"]))
                if layer1["name"] != layer2["name"]:
                    raise IncompatibleStateError(
                        "Neural network state has {0}={1}, while this architecture "
                        "has {0}={2}.".format("name", layer2["name"], layer1["name"]))
                if layer1["output"] != layer2["output"]:
                    raise IncompatibleStateError(
                        "Neural network state has {0}={1}, while this architecture "
                        "has {0}={2}.".format("output", layer2["output"], layer1["output"]))

    def __init__(self, dictionary, architecture, batch_processing=True, profile=False):
        Initializes the neural network parameters for all layers, and
        creates Theano shared variables from them.

After Change

            for layer_id, layer in enumerate(self.layers):
                h5_layer = h5_layers[str(layer_id)]
                for variable, values in layer.items():
                    if isinstance(values, list):
                        h5_values = h5_layer[variable]
                        for value_id, value in enumerate(values):
                            h5_value = h5_values.attrs[str(value_id)]
                            if value != h5_value:
                                raise IncompatibleStateError(
                                    "Neural network state has {0}={1}, while "
                                    "this architecture has {0}={2}.".format(
                                        variable, value, h5_value))
                    else:
                        h5_value = h5_layer.attrs[variable]
                        if values != h5_value:
                            raise IncompatibleStateError(
                                "Neural network state has {0}={1}, while "
                                "this architecture has {0}={2}.".format(
                                    variable, value, h5_value))

    def __init__(self, dictionary, architecture, batch_processing=True, profile=False):
        Initializes the neural network parameters for all layers, and
        creates Theano shared variables from them.

Instances