nnproto

package

v0.6.17 Latest Latest Go to latest Published: Aug 22, 2026 License: MIT Imports: 3 Imported by: 0

Details

Valid go.mod file
Redistributable license
Tagged version
Stable version
Learn more about best practices

Repository

github.com/tmc/apple

Links

Open Source Insights

Documentation ¶

Overview ¶

Package nnproto encodes CoreML NeuralNetwork models (specification version 1) as .mlmodel protobuf bytes. It provides the protobuf construction required to build ANE-executable models without external dependencies.

Index ¶

func FFNReLU2(d, f, seq int, w1, w2 []float32) ([]byte, error)
type Feature
type Layer
type Model
- func (m Model) Encode() ([]byte, error)

Constants ¶

This section is empty.

Variables ¶

This section is empty.

Functions ¶

func FFNReLU2 ¶

func FFNReLU2(d, f, seq int, w1, w2 []float32) ([]byte, error)

FFNReLU2 builds a Model computing out = W2·relu(W1·x)² over input [d, 1, seq] with argument validation.

Types ¶

type Feature ¶

type Feature struct {
	Name  string
	Shape []int64
}

Feature names a model input or output and specifies its shape.

type Layer ¶

type Layer []byte

Layer is an encoded NeuralNetworkLayer message.

func Conv1x1 ¶

func Conv1x1(name, input, output string, inC, outC int, weights []float32) (Layer, error)

Conv1x1 encodes a 1x1 Convolution NeuralNetworkLayer with argument validation.

func Multiply ¶

func Multiply(name string, inputs []string, output string) Layer

Multiply encodes an elementwise multiply NeuralNetworkLayer.

func ReLU ¶

func ReLU(name, input, output string) Layer

ReLU encodes a ReLU activation NeuralNetworkLayer.

type Model ¶

type Model struct {
	Inputs  []Feature
	Outputs []Feature
	Layers  []Layer
}

Model describes a CoreML NeuralNetwork model.

func (Model) Encode ¶

func (m Model) Encode() ([]byte, error)

Encode serializes the Model into .mlmodel protobuf bytes with validation.

Source Files ¶

View all Source files

nnproto.go

?	: This menu
/	: Search site
f or F	: Jump to
y or Y	: Canonical URL