LayerAttributionResult<T>

Models & Types · AiDotNet.Interpretability.Explainers

Result of layer attribution.

Constructors

Constructor Summary
LayerAttributionResult()

Properties

Property Summary
LayerAttributionMethod Method { get; set; } Gets or sets the attribution method used.
Vector<T> Activations { get; set; } Gets or sets the layer activations.
Vector<T> Attributions { get; set; } Gets or sets attribution scores for each neuron in the layer.
Vector<T> Instance { get; set; } Gets or sets the input instance.
Vector<T> Prediction { get; set; } Gets or sets the model prediction.
int TargetClass { get; set; } Gets or sets the target class.

Methods

Method Summary
IEnumerable<ValueTuple<int, T, T>> GetTopNeurons(int k) Gets the top K neurons by attribution magnitude.
T GetAttributionSum() Gets the sum of attributions.
Tensor<T> ReshapeSpatial(int height, int width, int channels) Reshapes attributions for spatial layers (e.g., conv layers).
Vector<T> GetChannelAttributions(int height, int width, int channels) Gets channel-wise attribution sums (for conv layers).
string ToString() Returns a human-readable summary.