class InputXBaselineGradientExplainer(FeatureAttributionExplainer):
    """Input × Baseline Gradient explainer for computing scaled baseline-gradient attributions.

    This explainer computes attributions by multiplying (input - baseline) with their
    gradients, providing a measure that considers both the deviation from baseline
    and gradient sensitivity. This method is particularly useful when you have
    meaningful baseline references. Supports both single-target and multi-target
    modes for both single-target and multi-target scenarios.

    The Input × Baseline Gradient method provides attributions that are grounded
    in both the input magnitude relative to a baseline and gradient information.

    Args:
        model: The PyTorch model whose output is to be explained.
        multi_target: Whether to use multi-target mode. When True, can compute
            attributions for multiple targets simultaneously. Defaults to False.
        internal_batch_size: Batch size for internal computations. Defaults to 64.
        grad_batch_size: Batch size for gradient computations. Defaults to 64.

    Examples:
        Single-target usage:
        >>> import torch
        >>> from torchxai.data_types import SingleTargetAcrossBatch
        >>>
        >>> model = torch.nn.Linear(10, 2)
        >>> explainer = InputXBaselineGradientExplainer(model)
        >>> inputs   = torch.randn(1, 10)
        >>> baseline = torch.zeros(1, 10)
        >>> attributions = explainer.explain(
        ...     inputs=inputs,
        ...     baselines=baseline,
        ...     target=SingleTargetAcrossBatch(index=0),
        ... )
        >>> attributions.shape   # (1, 10)

        Multi-target usage:
        >>> explainer_mt = InputXBaselineGradientExplainer(model, multi_target=True)
        >>> mt_attributions = explainer_mt.explain(
        ...     inputs=inputs,
        ...     baselines=baseline,
        ...     target=[SingleTargetAcrossBatch(index=0), SingleTargetAcrossBatch(index=1)],
        ... )
        >>> len(mt_attributions), mt_attributions[0].shape   # 2, (1, 10)
    """

    def _init_single_target_explanation_fn(self) -> Callable:
        """Initialize single-target Input × Gradient attribution function.

        Returns:
            Captum InputXGradient attribution function for single targets.
        """
        return InputBaselineXGradient(self._model).attribute

    def _init_multi_target_explanation_fn(self) -> Callable:
        """Initialize multi-target Input × Gradient attribution function.

        Returns:
            MultiTargetInputXGradient attribution function for multiple targets.
        """
        return MultiTargetInputBaselineXGradient(
            self._model, grad_batch_size=self._grad_batch_size
        ).attribute

    def explain(
        self,
        inputs: TensorOrTupleOfTensorsGeneric,
        target: ExplanationTargetType | list[ExplanationTargetType],
        baselines: TensorOrTupleOfTensorsGeneric | None = None,
        additional_forward_args: tuple[Any, ...] | None = None,
    ) -> TensorOrTupleOfTensorsGeneric | list[TensorOrTupleOfTensorsGeneric]:
        """Compute Input × Baseline Gradient attributions for the given inputs.

        Args:
            inputs: Input tensor(s) for attribution computation.
            target: An `ExplanationTargetType` (e.g. `SingleTargetAcrossBatch`) for single-target
                mode, or a list of them for multi-target mode.
            baselines: Baseline tensors representing reference values. Required for
                this method as it computes (input - baseline) × gradient.
            additional_forward_args: Additional arguments for model forward pass.

        Returns:
            Tensor in single-target mode. List of Tensors, one per target, in multi-target mode.

        Examples:
            >>> # With explicit baselines
            >>> attributions = explainer.explain(
            ...     inputs=OrderedDict({"input": torch.randn(2, 10)}),
            ...     target=torch.tensor([0, 1]),
            ...     baselines=OrderedDict({"input": torch.zeros(2, 10)}),
            ... )
            >>>
            >>> # Multiple features with baselines
            >>> attributions = explainer.explain(
            ...     inputs=OrderedDict(
            ...         {"feat1": torch.randn(2, 5), "feat2": torch.randn(2, 5)}
            ...     ),
            ...     target=torch.tensor([0, 1]),
            ...     baselines=OrderedDict(
            ...         {"feat1": torch.zeros(2, 5), "feat2": torch.zeros(2, 5)}
            ...     ),
            ... )
        """
        return self._default_explain(
            inputs=inputs,
            target=target,
            baselines=baselines,
            additional_forward_args=additional_forward_args,
        )