skainet-compile-core/sk.ainet.compile.nn

Package-level declarations

Types

DefaultNetworkContext

class DefaultNetworkContext<T : DType, V> : NeuralNetworkContext<T, V>

DefaultOperationRegistry

class DefaultOperationRegistry : OperationRegistry

Default implementation of OperationRegistry

GlobalOperationRegistry

object GlobalOperationRegistry

Global operation registry

NeuralNetworkContext

interface NeuralNetworkContext<T : DType, V>

Context for the DSL to define the data type and operations.

OperationFactory

interface OperationFactory

Factory interface for creating operations

OperationMetadata

data class OperationMetadata(val name: String, val type: String, val description: String, val inputSpecs: List<ParameterSpec>, val outputSpecs: List<ParameterSpec>, val parameterSpecs: List<ParameterSpec>, val supportsGradients: Boolean = false, val additionalMetadata: Map<String, Any> = emptyMap())

Metadata about an operation type

OperationRegistry

interface OperationRegistry

Registry for managing operation types and their implementations. Provides centralized management of available operations and their metadata.

ParameterSpec

data class ParameterSpec(val name: String, val type: String, val required: Boolean = true, val defaultValue: Any? = null, val description: String = "", val constraints: Map<String, Any> = emptyMap())

Specification for a parameter (input, output, or configuration parameter)

Functions

definition

fun <T : DType, V> definition(init: NeuralNetworkContext<T, V>.(NeuralNetworkContext<T, V>) -> Module<T, V>): Module<T, V>

Creates a context for the DSL with the given configuration.

network

inline fun <T : DType, V> NeuralNetworkContext<T, V>.network(content: NeuralNetworkDsl<T, V>.() -> Unit): Module<T, V>

Extension function to create a network within a NetworkContext. This bridges the context wrapper with the network DSL using the context's tensor factory.