xor

Documentation

Overview

Package xor implements a extremely simple network that can perform the xor operation.

Index

• type Benchmark
  • func NewBenchmark(driver *driver.Driver) *Benchmark
  • func (b *Benchmark) Run()
  • func (b *Benchmark) SelectGPU(gpuIDs []int)
  • func (b *Benchmark) SetUnifiedMemory()
  • func (b *Benchmark) Verify()
• type DataSource
  • func NewDataSource(to tensor.Operator) *DataSource
  • func (ds *DataSource) NextBatch(batchSize int) (data tensor.Tensor, label []int)
  • func (ds *DataSource) Rewind()

Types

type Benchmark

type Benchmark struct {
	// contains filtered or unexported fields
}

Benchmark defines the XOR network training benchmark.

func NewBenchmark

func NewBenchmark(driver *driver.Driver) *Benchmark

NewBenchmark creates a new benchmark.

func (*Benchmark) Run

func (b *Benchmark) Run()

Run executes the benchmark.

func (*Benchmark) SelectGPU

func (b *Benchmark) SelectGPU(gpuIDs []int)

SelectGPU selects the GPU to use.

func (*Benchmark) SetUnifiedMemory

func (b *Benchmark) SetUnifiedMemory()

SetUnifiedMemory asks the benchmark to use unified memory.

func (*Benchmark) Verify

func (b *Benchmark) Verify()

Verify runs the benchmark on the CPU and checks the result.

type DataSource

type DataSource struct {
	// contains filtered or unexported fields
}

DataSource defines the training dataset for the xor operation.

func NewDataSource

func NewDataSource(to tensor.Operator) *DataSource

NewDataSource creates a new XOR datasource

func (*DataSource) NextBatch

func (ds *DataSource) NextBatch(batchSize int) (
	data tensor.Tensor,
	label []int,
)

NextBatch returns the next batch data.

func (*DataSource) Rewind

func (ds *DataSource) Rewind()

Rewind moves the pointer to the beginning of the training set.