skainet-lang-core/sk.ainet.lang.nn/Functional

Functional

Minimal multi-input functional wrapper to express a computation as a single callable model.

This is intentionally lightweight: it stores input metadata and a build lambda that performs the full forward pass using existing Modules/DSL and returns the final Tensor.

Types

Args

class Args<T : DType, V>(feed: Map<String, Tensor<T, V>>)

Arguments accessor passed to the compute lambda. Retrieves tensors by input name.

Companion

object Companion

Properties

inputs

val inputs: List<FuncInput>

Functions

forward

fun forward(feed: Map<String, Tensor<T, V>>, ctx: ExecutionContext): Tensor<T, V>

Executes the functional model with a map of named input tensors.