Overview > klyn.math.nn > Sequential > compile
Version 0.2.2 Overview Index Tree Deprecated
klyn.math.nn.Sequential.compile
method
public compile(optimizer as Optimizer<T>, loss as ClassificationLoss<T>, metrics as IList<String> = ["accuracy"]) as Void
Description

Configures the optimizer and mean classification loss without resetting weights.

Parameters
ParameterDescription
optimizerParameter update rule.
lossLoss consuming the final layer outputs and integer targets.
metricsCurrently [accuracy]; unknown metrics are rejected.
Throws
  • ValueException for null components or unsupported metrics.
Example

model.compile(SGD<Float>(0.1), SparseCategoricalCrossentropy<Float>())