@hoff97/tensor-js/dist/lib/ops/gpu/conv/normalize.d.ts

PyTorch like deep learning inferrence library

import { DTypeGpu, GPUTensorConstructor, GPUTensorI } from '../../../tensor/gpu/interface';
import { GPUMemoryAllocator } from '../../../tensor/gpu/memory';
import { Input, Operation } from '../operation';
export interface NormalizeOpInfo {
    shapeX?: readonly number[];
    widthX?: number;
    heightX?: number;
    shapeMean?: readonly number[];
    widthMean?: number;
    heightMean?: number;
    shapeVariance?: readonly number[];
    widthVariance?: number;
    heightVariance?: number;
    shapeScale?: readonly number[];
    widthScale?: number;
    heightScale?: number;
    shapeBias?: readonly number[];
    widthBias?: number;
    heightBias?: number;
    shapeOutput?: readonly number[];
    widthOutput?: number;
    heightOutput?: number;
    epsilon?: number;
}
export interface NormalizeOpInput {
    X: GPUTensorI;
    Mean: GPUTensorI;
    Variance: GPUTensorI;
    epsilon: number;
    Scale: GPUTensorI;
    Bias: GPUTensorI;
}
export declare class NormalizeOperation<GPUTensor extends GPUTensorI> extends Operation<GPUTensor, NormalizeOpInfo, NormalizeOpInput> {
    constructor(tensorConstructor: GPUTensorConstructor<GPUTensor>, dtype: DTypeGpu, allocator?: GPUMemoryAllocator);
    getVariables(): string;
    getUniformAttrs(): Input[];
    getFragmentShader(info: NormalizeOpInfo): string;
    getOutputShape(input: NormalizeOpInput): readonly number[];
    getTextureNames(): string[];
    calc(input: NormalizeOpInput): GPUTensor;
    compile(info: NormalizeOpInfo): void;
    getCompilationInfo(input: NormalizeOpInput): NormalizeOpInfo;
    getInputInfoString(input: NormalizeOpInput): string;
}