Class GradSpMMOp

Namespace

AiDotNet.JitCompiler.IR.Operations

Assembly

AiDotNet.dll

Represents the gradient of sparse matrix-matrix multiplication in the IR.

public class GradSpMMOp : IROp

Inheritance

object

IROp

GradSpMMOp

Inherited Members

IROp.OutputIds

IROp.OutputId

IROp.InputIds

IROp.OutputType

IROp.OutputShape

IROp.OutputShapes

IROp.OpType

IROp.ToString()

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

Remarks

For SpMM C = A * B: - Gradient w.r.t. B: d_B = A^T * d_C - Gradient w.r.t. A (sparse): d_A[i,j] = sum_k(d_C[i,k] * B[j,k]) for non-zero positions

Properties

DenseColumns

Gets or sets the number of columns in the dense matrix.

public int DenseColumns { get; set; }

Property Value

int

SparseColumns

Gets or sets the number of columns in the sparse matrix.

public int SparseColumns { get; set; }

Property Value

int

SparseRows

Gets or sets the number of rows in the sparse matrix.

public int SparseRows { get; set; }

Property Value

int

Methods

Validate()

Validates that this operation is correctly formed.

public override bool Validate()

Returns

bool

True if valid, false otherwise.