2de1df1704c1c187c72aacc15efcacd7029139dd,stellargraph/utils/saliency_maps/integrated_gradients.py,IntegratedGradients,__init__,#IntegratedGradients#Any#Any#,40

Before Change

    

    def __init__(self, model, generator):
        super().__init__(model, generator)

    def get_integrated_node_masks(
        self,
        node_idx,

After Change

        // Set sparse flag from the generator
        self._is_sparse = generator.use_sparse

        if self._is_sparse:
            if not isinstance(generator, SparseFullBatchSequence):
                raise TypeError(
                    "The generator supplied has to be an object of SparseFullBatchSequence for sparse adjacency matrix."
                )
            if len(model.input) != 4:
                raise RuntimeError(
                    "Keras model for sparse adjacency is expected to have four inputs"
                )
            self._adj = generator.A_values
            self._adj_inds = generator.A_indices
        else:
            if not isinstance(generator, FullBatchSequence):
                raise TypeError(
                    "The generator supplied has to be an object of FullBatchSequence for dense adjacency matrix."
                )
            if len(model.input) != 3:
                raise RuntimeError(
                    "Keras model for dense adjacency is expected to have three inputs"
                )

            self._adj = generator.A_dense

        // Extract features from generator
        self._features = generator.features
        self._model = model

    def get_integrated_node_masks(

Instances