Class Tensor<R>

A tf.Tensor object represents an immutable, multidimensional array of numbers that has a shape and a data type.

For performance reasons, functions that create tensors do not necessarily perform a copy of the data passed to them (e.g. if the data is passed as a Float32Array), and changes to the data will change the tensor. This is not a feature and is not supported. To avoid this behavior, use the tensor before changing the input data or create a copy with copy = tf.add(yourTensor, 0).

See tf.tensor for details on how to create a tf.Tensor.

Doc

Type Parameters

Hierarchy (View Summary)

Implements

  • TensorInfo

Constructors

constructor

Properties

dataId dtype id kept kerasMask? rankType scopeId shape size strides

Accessors

isDisposed rank

Methods

abs acos acosh add all any argMax argMin array arraySync as1D as2D as3D as4D as5D asin asinh asScalar asType atan atan2 atanh avgPool batchNorm batchToSpaceND broadcastTo buffer bufferSync bytes cast ceil clipByValue clone concat conv1d conv2d conv2dTranspose cos cosh cumprod cumsum data dataSync dataToGPU depthToSpace depthwiseConv2d dilation2d dispose div divNoNan dot elu equal erf euclideanNorm exp expandDims expm1 fft flatten floor floorDiv gather greater greaterEqual ifft irfft isFinite isInf isNaN leakyRelu less lessEqual localResponseNormalization log log1p logicalAnd logicalNot logicalOr logicalXor logSigmoid logSoftmax logSumExp matMul max maximum maxPool mean min minimum mirrorPad mod mul neg norm notEqual oneHot onesLike pad pool pow prelu print prod reciprocal relu relu6 reshape reshapeAs resizeBilinear resizeNearestNeighbor reverse rfft round rsqrt selu separableConv2d sigmoid sign sin sinh slice softmax softplus spaceToBatchND split sqrt square squaredDifference squeeze stack step stridedSlice sub sum tan tanh throwIfDisposed tile toBool toFloat toInt topk toString transpose unique unsortedSegmentSum unstack variable where zerosLike

Constructors

constructor

  • new Tensor<R extends Rank = Rank>(
        shape: ShapeMap[R],
        dtype: keyof DataTypeMap,
        dataId: object,
        id: number,
    ): tf.Tensor<R>

  • Type Parameters

    Parameters

    • shape: ShapeMap[R]
    • dtype: keyof DataTypeMap
    • dataId: object
    • id: number

    Returns tf.Tensor<R>

Properties

dataId

dataId: object

Id of the bucket holding the data for this tensor. Multiple arrays can point to the same bucket (e.g. when calling array.reshape()).

Readonlydtype

dtype: keyof DataTypeMap

The data type for the array.

Readonlyid

id: number

Unique id of this tensor.

kept

kept: boolean

Whether this tensor has been globally kept.

OptionalkerasMask

kerasMask?: tf.Tensor

The keras mask that some keras layers attach to the tensor

ReadonlyrankType

rankType: R

The rank type for the array (see Rank enum).

scopeId

scopeId: number

The id of the scope this tensor is being tracked in.

Readonlyshape

shape: ShapeMap[R]

The shape of the tensor.

Readonlysize

size: number

Number of elements in the tensor.

Readonlystrides

strides: number[]

Number of elements to skip in each dimension when indexing. See https://docs.scipy.org/doc/numpy/reference/generated/\ numpy.ndarray.strides.html

Accessors

isDisposed

  • get isDisposed(): boolean

  • Returns boolean

rank

  • get rank(): number

  • Returns number

Methods

abs

  • abs<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

acos

  • acos<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

acosh

  • acosh<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

add

all

  • all<T extends tf.Tensor>(
        this: T,
        axis?: number | number[],
        keepDims?: boolean,
    ): T

  • Type Parameters

    Parameters

    • this: T
    • Optionalaxis: number | number[]
    • OptionalkeepDims: boolean

    Returns T

any

  • any<T extends tf.Tensor>(
        this: T,
        axis?: number | number[],
        keepDims?: boolean,
    ): T

  • Type Parameters

    Parameters

    • this: T
    • Optionalaxis: number | number[]
    • OptionalkeepDims: boolean

    Returns T

argMax

  • argMax<T extends tf.Tensor>(axis?: number): T

  • Type Parameters

    Parameters

    • Optionalaxis: number

    Returns T

argMin

  • argMin<T extends tf.Tensor>(axis?: number): T

  • Type Parameters

    Parameters

    • Optionalaxis: number

    Returns T

array

  • array(): Promise<ArrayMap[R]>

  • Returns the tensor data as a nested array. The transfer of data is done asynchronously.

    Returns Promise<ArrayMap[R]>

    Doc

arraySync

  • arraySync(): ArrayMap[R]

  • Returns the tensor data as a nested array. The transfer of data is done synchronously.

    Returns ArrayMap[R]

    Doc

as1D

  • as1D<T extends tf.Tensor>(): Tensor1D

  • Type Parameters

    Returns Tensor1D

as2D

  • as2D<T extends tf.Tensor>(rows: number, columns: number): Tensor2D

  • Type Parameters

    Parameters

    • rows: number
    • columns: number

    Returns Tensor2D

as3D

  • as3D<T extends tf.Tensor>(
        rows: number,
        columns: number,
        depth: number,
    ): Tensor3D

  • Type Parameters

    Parameters

    • rows: number
    • columns: number
    • depth: number

    Returns Tensor3D

as4D

  • as4D<T extends tf.Tensor>(
        rows: number,
        columns: number,
        depth: number,
        depth2: number,
    ): Tensor4D

  • Type Parameters

    Parameters

    • rows: number
    • columns: number
    • depth: number
    • depth2: number

    Returns Tensor4D

as5D

  • as5D<T extends tf.Tensor>(
        rows: number,
        columns: number,
        depth: number,
        depth2: number,
        depth3: number,
    ): Tensor5D

  • Type Parameters

    Parameters

    • rows: number
    • columns: number
    • depth: number
    • depth2: number
    • depth3: number

    Returns Tensor5D

asin

  • asin<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

asinh

  • asinh<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

asScalar

  • asScalar<T extends tf.Tensor>(): Scalar

  • Type Parameters

    Returns Scalar

asType

  • asType<T extends tf.Tensor>(this: T, dtype: keyof DataTypeMap): T

  • Type Parameters

    Parameters

    • this: T
    • dtype: keyof DataTypeMap

    Returns T

atan

  • atan<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

atan2

atanh

  • atanh<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

avgPool

  • avgPool<T extends Tensor4D | Tensor3D>(
        filterSize: number | [number, number],
        strides: number | [number, number],
        pad: number | "valid" | "same" | ExplicitPadding,
        dimRoundingMode?: "floor" | "round" | "ceil",
    ): T

  • Type Parameters

    Parameters

    • filterSize: number | [number, number]
    • strides: number | [number, number]
    • pad: number | "valid" | "same" | ExplicitPadding
    • OptionaldimRoundingMode: "floor" | "round" | "ceil"

    Returns T

batchNorm

batchToSpaceND

  • batchToSpaceND<R extends Rank>(
        blockShape: number[],
        crops: number[][],
    ): tf.Tensor<R>

  • Type Parameters

    Parameters

    • blockShape: number[]
    • crops: number[][]

    Returns tf.Tensor<R>

broadcastTo

  • broadcastTo<R extends Rank>(shape: ShapeMap[R]): tf.Tensor<R>

  • Type Parameters

    Parameters

    • shape: ShapeMap[R]

    Returns tf.Tensor<R>

buffer

  • buffer<D extends keyof DataTypeMap = "float32">(): Promise<TensorBuffer<R, D>>

  • Returns a promise of tf.TensorBuffer that holds the underlying data.

    Type Parameters

    • D extends keyof DataTypeMap = "float32"

    Returns Promise<TensorBuffer<R, D>>

    Doc

bufferSync

  • bufferSync<D extends keyof DataTypeMap = "float32">(): TensorBuffer<R, D>

  • Returns a tf.TensorBuffer that holds the underlying data.

    Type Parameters

    • D extends keyof DataTypeMap = "float32"

    Returns TensorBuffer<R, D>

    Doc

bytes

  • bytes(): Promise<Uint8Array | Uint8Array[]>

  • Returns the underlying bytes of the tensor's data.

    Returns Promise<Uint8Array | Uint8Array[]>

cast

  • cast<T extends tf.Tensor>(dtype: keyof DataTypeMap): T

  • Type Parameters

    Parameters

    • dtype: keyof DataTypeMap

    Returns T

ceil

  • ceil<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

clipByValue

  • clipByValue<T extends tf.Tensor>(min: number, max: number): tf.Tensor

  • Type Parameters

    Parameters

    • min: number
    • max: number

    Returns tf.Tensor

clone

  • clone<T extends tf.Tensor>(this: T): T

  • Returns a copy of the tensor. See tf.clone for details.

    Type Parameters

    Parameters

    • this: T

    Returns T

    Doc

concat

  • concat<T extends tf.Tensor>(tensors: T | (TensorLike | T)[], axis?: number): T

  • Type Parameters

    Parameters

    Returns T

conv1d

  • conv1d<T extends Tensor2D | Tensor3D>(
        filter: Tensor3D | TensorLike3D,
        stride: number,
        pad: number | "valid" | "same" | ExplicitPadding,
        dataFormat?: "NWC" | "NCW",
        dilation?: number,
        dimRoundingMode?: "floor" | "round" | "ceil",
    ): T

  • Type Parameters

    Parameters

    • filter: Tensor3D | TensorLike3D
    • stride: number
    • pad: number | "valid" | "same" | ExplicitPadding
    • OptionaldataFormat: "NWC" | "NCW"
    • Optionaldilation: number
    • OptionaldimRoundingMode: "floor" | "round" | "ceil"

    Returns T

conv2d

  • conv2d<T extends Tensor4D | Tensor3D>(
        filter: Tensor4D | TensorLike4D,
        strides: number | [number, number],
        pad: number | "valid" | "same",
        dataFormat?: "NHWC" | "NCHW",
        dilations?: number | [number, number],
        dimRoundingMode?: "floor" | "round" | "ceil",
    ): T

  • Type Parameters

    Parameters

    • filter: Tensor4D | TensorLike4D
    • strides: number | [number, number]
    • pad: number | "valid" | "same"
    • OptionaldataFormat: "NHWC" | "NCHW"
    • Optionaldilations: number | [number, number]
    • OptionaldimRoundingMode: "floor" | "round" | "ceil"

    Returns T

conv2dTranspose

  • conv2dTranspose<T extends Tensor4D | Tensor3D>(
        filter: Tensor4D | TensorLike4D,
        outputShape: [number, number, number, number] | [number, number, number],
        strides: number | [number, number],
        pad: number | "valid" | "same",
        dimRoundingMode?: "floor" | "round" | "ceil",
    ): T

  • Type Parameters

    Parameters

    • filter: Tensor4D | TensorLike4D
    • outputShape: [number, number, number, number] | [number, number, number]
    • strides: number | [number, number]
    • pad: number | "valid" | "same"
    • OptionaldimRoundingMode: "floor" | "round" | "ceil"

    Returns T

cos

  • cos<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

cosh

  • cosh<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

cumprod

  • cumprod<R extends Rank>(
        axis?: number,
        exclusive?: boolean,
        reverse?: boolean,
    ): tf.Tensor<R>

  • Type Parameters

    Parameters

    • Optionalaxis: number
    • Optionalexclusive: boolean
    • Optionalreverse: boolean

    Returns tf.Tensor<R>

cumsum

  • cumsum<R extends Rank>(
        axis?: number,
        exclusive?: boolean,
        reverse?: boolean,
    ): tf.Tensor<R>

  • Type Parameters

    Parameters

    • Optionalaxis: number
    • Optionalexclusive: boolean
    • Optionalreverse: boolean

    Returns tf.Tensor<R>

data

  • data<D extends keyof DataTypeMap = NumericDataType>(): Promise<DataTypeMap[D]>

  • Asynchronously downloads the values from the tf.Tensor. Returns a promise of TypedArray that resolves when the computation has finished.

    Type Parameters

    • D extends keyof DataTypeMap = NumericDataType

    Returns Promise<DataTypeMap[D]>

    Doc

dataSync

  • dataSync<D extends keyof DataTypeMap = NumericDataType>(): DataTypeMap[D]

  • Synchronously downloads the values from the tf.Tensor. This blocks the UI thread until the values are ready, which can cause performance issues.

    Type Parameters

    • D extends keyof DataTypeMap = NumericDataType

    Returns DataTypeMap[D]

    Doc

dataToGPU

  • dataToGPU(options?: DataToGPUWebGLOption): GPUData

  • Copy the tensor's data to a new GPU resource. Comparing to the dataSync() and data(), this method prevents data from being downloaded to CPU.

    For WebGL backend, the data will be stored on a densely packed texture. This means that the texture will use the RGBA channels to store value.

    For WebGPU backend, the data will be stored on a buffer. There is no parameter, so can not use a user-defined size to create the buffer.

    Parameters

    • Optionaloptions: DataToGPUWebGLOption

    Returns GPUData

    For WebGL backend, a GPUData contains the new texture and its information. { tensorRef: The tensor that is associated with this texture, texture: WebGLTexture, texShape: [number, number] // [height, width] }

    For WebGPU backend, a GPUData contains the new buffer.
{
   tensorRef: The tensor that is associated with this buffer,
   buffer: GPUBuffer,
}

Remember to dispose the GPUData after it is used by
`res.tensorRef.dispose()`.

    Doc

depthToSpace

  • depthToSpace<T extends Tensor4D>(
        blockSize: number,
        dataFormat: "NHWC" | "NCHW",
    ): T

  • Type Parameters

    Parameters

    • blockSize: number
    • dataFormat: "NHWC" | "NCHW"

    Returns T

depthwiseConv2d

  • depthwiseConv2d<T extends Tensor4D | Tensor3D>(
        filter: Tensor4D | TensorLike4D,
        strides: number | [number, number],
        pad: number | "valid" | "same",
        dataFormat?: "NHWC" | "NCHW",
        dilations?: number | [number, number],
        dimRoundingMode?: "floor" | "round" | "ceil",
    ): T

  • Type Parameters

    Parameters

    • filter: Tensor4D | TensorLike4D
    • strides: number | [number, number]
    • pad: number | "valid" | "same"
    • OptionaldataFormat: "NHWC" | "NCHW"
    • Optionaldilations: number | [number, number]
    • OptionaldimRoundingMode: "floor" | "round" | "ceil"

    Returns T

dilation2d

  • dilation2d<T extends Tensor4D | Tensor3D>(
        filter: Tensor3D | TensorLike3D,
        strides: number | [number, number],
        pad: "valid" | "same",
        dilations?: number | [number, number],
        dataFormat?: "NHWC",
    ): T

  • Type Parameters

    Parameters

    • filter: Tensor3D | TensorLike3D
    • strides: number | [number, number]
    • pad: "valid" | "same"
    • Optionaldilations: number | [number, number]
    • OptionaldataFormat: "NHWC"

    Returns T

dispose

  • dispose(): void

  • Disposes tf.Tensor from memory.

    Returns void

    Doc

div

divNoNan

dot

elu

  • elu<T extends tf.Tensor>(): T

  • Type Parameters

    Returns T

equal

erf

  • erf<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

euclideanNorm

  • euclideanNorm<T extends tf.Tensor>(
        this: T,
        axis?: number | number[],
        keepDims?: boolean,
    ): T

  • Type Parameters

    Parameters

    • this: T
    • Optionalaxis: number | number[]
    • OptionalkeepDims: boolean

    Returns T

exp

  • exp<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

expandDims

  • expandDims<T extends tf.Tensor>(axis?: number): T

  • Type Parameters

    Parameters

    • Optionalaxis: number

    Returns T

expm1

  • expm1<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

fft

  • fft<T extends tf.Tensor>(this: tf.Tensor): tf.Tensor

  • Type Parameters

    Parameters

    Returns tf.Tensor

flatten

  • flatten<T extends tf.Tensor>(): Tensor1D

  • Type Parameters

    Returns Tensor1D

floor

  • floor<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

floorDiv

gather

  • gather<T extends tf.Tensor>(
        this: T,
        indices: TensorLike | tf.Tensor,
        axis?: number,
        batchDims?: number,
    ): T

  • Type Parameters

    Parameters

    Returns T

greater

greaterEqual

  • greaterEqual<T extends tf.Tensor>(b: TensorLike | tf.Tensor): T

  • Type Parameters

    Parameters

    Returns T

ifft

  • ifft<T extends tf.Tensor>(this: tf.Tensor): tf.Tensor

  • Type Parameters

    Parameters

    Returns tf.Tensor

irfft

  • irfft<T extends tf.Tensor>(this: tf.Tensor): tf.Tensor

  • Type Parameters

    Parameters

    Returns tf.Tensor

isFinite

  • isFinite<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

isInf

  • isInf<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

isNaN

  • isNaN<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

leakyRelu

  • leakyRelu<T extends tf.Tensor>(alpha: number): T

  • Type Parameters

    Parameters

    • alpha: number

    Returns T

less

lessEqual

localResponseNormalization

  • localResponseNormalization<T extends tf.Tensor>(
        depthRadius?: number,
        bias?: number,
        alpha?: number,
        beta?: number,
    ): T

  • Type Parameters

    Parameters

    • OptionaldepthRadius: number
    • Optionalbias: number
    • Optionalalpha: number
    • Optionalbeta: number

    Returns T

log

  • log<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

log1p

  • log1p<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

logicalAnd

  • logicalAnd<T extends tf.Tensor>(b: TensorLike | tf.Tensor): T

  • Type Parameters

    Parameters

    Returns T

logicalNot

  • logicalNot<T extends tf.Tensor>(): T

  • Type Parameters

    Returns T

logicalOr

logicalXor

  • logicalXor<T extends tf.Tensor>(b: TensorLike | tf.Tensor): T

  • Type Parameters

    Parameters

    Returns T

logSigmoid

  • logSigmoid<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

logSoftmax

  • logSoftmax<T extends tf.Tensor>(this: T, axis?: number): T

  • Type Parameters

    Parameters

    • this: T
    • Optionalaxis: number

    Returns T

logSumExp

  • logSumExp<T extends tf.Tensor>(
        this: T,
        axis?: number | number[],
        keepDims?: boolean,
    ): T

  • Type Parameters

    Parameters

    • this: T
    • Optionalaxis: number | number[]
    • OptionalkeepDims: boolean

    Returns T

matMul

  • matMul<T extends tf.Tensor>(
        b: TensorLike | tf.Tensor,
        transposeA?: boolean,
        transposeB?: boolean,
    ): tf.Tensor

  • Type Parameters

    Parameters

    Returns tf.Tensor

max

  • max<T extends tf.Tensor>(axis?: number | number[], keepDims?: boolean): T

  • Type Parameters

    Parameters

    • Optionalaxis: number | number[]
    • OptionalkeepDims: boolean

    Returns T

maximum

maxPool

  • maxPool<T extends Tensor4D | Tensor3D>(
        filterSize: number | [number, number],
        strides: number | [number, number],
        pad: number | "valid" | "same" | ExplicitPadding,
        dimRoundingMode?: "floor" | "round" | "ceil",
    ): T

  • Type Parameters

    Parameters

    • filterSize: number | [number, number]
    • strides: number | [number, number]
    • pad: number | "valid" | "same" | ExplicitPadding
    • OptionaldimRoundingMode: "floor" | "round" | "ceil"

    Returns T

mean

  • mean<T extends tf.Tensor>(axis?: number | number[], keepDims?: boolean): T

  • Type Parameters

    Parameters

    • Optionalaxis: number | number[]
    • OptionalkeepDims: boolean

    Returns T

min

  • min<T extends tf.Tensor>(axis?: number | number[], keepDims?: boolean): T

  • Type Parameters

    Parameters

    • Optionalaxis: number | number[]
    • OptionalkeepDims: boolean

    Returns T

minimum

mirrorPad

  • mirrorPad<T extends tf.Tensor>(
        paddings: [number, number][],
        mode: "reflect" | "symmetric",
    ): T

  • Type Parameters

    Parameters

    • paddings: [number, number][]
    • mode: "reflect" | "symmetric"

    Returns T

mod

mul

neg

  • neg<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

norm

  • norm<T extends tf.Tensor>(
        ord?: number | "euclidean" | "fro",
        axis?: number | number[],
        keepDims?: boolean,
    ): tf.Tensor

  • Type Parameters

    Parameters

    • Optionalord: number | "euclidean" | "fro"
    • Optionalaxis: number | number[]
    • OptionalkeepDims: boolean

    Returns tf.Tensor

notEqual

oneHot

  • oneHot(depth: number, onValue: number, offValue: number): tf.Tensor

  • Parameters

    • depth: number
    • onValue: number
    • offValue: number

    Returns tf.Tensor

onesLike

  • onesLike<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

pad

  • pad<T extends tf.Tensor>(
        paddings: [number, number][],
        constantValue?: number,
    ): T

  • Type Parameters

    Parameters

    • paddings: [number, number][]
    • OptionalconstantValue: number

    Returns T

pool

  • pool<T extends Tensor4D | Tensor3D>(
        windowShape: number | [number, number],
        poolingType: "avg" | "max",
        padding: number | "valid" | "same" | ExplicitPadding,
        diationRate?: number | [number, number],
        strides?: number | [number, number],
        dimRoundingMode?: "floor" | "round" | "ceil",
    ): T

  • Type Parameters

    Parameters

    • windowShape: number | [number, number]
    • poolingType: "avg" | "max"
    • padding: number | "valid" | "same" | ExplicitPadding
    • OptionaldiationRate: number | [number, number]
    • Optionalstrides: number | [number, number]
    • OptionaldimRoundingMode: "floor" | "round" | "ceil"

    Returns T

pow

prelu

  • prelu<T extends tf.Tensor>(alpha: TensorLike | T): T

  • Type Parameters

    Parameters

    Returns T

print

  • print(verbose?: boolean): void

  • Prints the tf.Tensor. See tf.print for details.

    Parameters

    • Optionalverbose: boolean

      Whether to print verbose information about the tensor, including dtype and size.

    Returns void

    Doc

prod

  • prod<T extends tf.Tensor>(
        this: T,
        axis?: number | number[],
        keepDims?: boolean,
    ): T

  • Type Parameters

    Parameters

    • this: T
    • Optionalaxis: number | number[]
    • OptionalkeepDims: boolean

    Returns T

reciprocal

  • reciprocal<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

relu

  • relu<T extends tf.Tensor>(): T

  • Type Parameters

    Returns T

relu6

  • relu6<T extends tf.Tensor>(): T

  • Type Parameters

    Returns T

reshape

  • reshape<T extends tf.Tensor>(shape: number[]): T

  • Type Parameters

    Parameters

    • shape: number[]

    Returns T

reshapeAs

  • reshapeAs<T extends tf.Tensor>(x: T): T

  • Type Parameters

    Parameters

    Returns T

resizeBilinear

  • resizeBilinear<T extends Tensor4D | Tensor3D>(
        newShape2D: [number, number],
        alignCorners?: boolean,
        halfPixelCenters?: boolean,
    ): T

  • Type Parameters

    Parameters

    • newShape2D: [number, number]
    • OptionalalignCorners: boolean
    • OptionalhalfPixelCenters: boolean

    Returns T

resizeNearestNeighbor

  • resizeNearestNeighbor<T extends Tensor4D | Tensor3D>(
        newShape2D: [number, number],
        alignCorners?: boolean,
        halfFloatCenters?: boolean,
    ): T

  • Type Parameters

    Parameters

    • newShape2D: [number, number]
    • OptionalalignCorners: boolean
    • OptionalhalfFloatCenters: boolean

    Returns T

reverse

  • reverse<T extends tf.Tensor>(this: T, axis?: number | number[]): T

  • Type Parameters

    Parameters

    • this: T
    • Optionalaxis: number | number[]

    Returns T

rfft

  • rfft<T extends tf.Tensor>(this: tf.Tensor): tf.Tensor

  • Type Parameters

    Parameters

    Returns tf.Tensor

round

  • round<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

rsqrt

  • rsqrt<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

selu

  • selu<T extends tf.Tensor>(): T

  • Type Parameters

    Returns T

separableConv2d

  • separableConv2d<T extends Tensor4D | Tensor3D>(
        depthwiseFilter: Tensor4D | TensorLike4D,
        pointwiseFilter: Tensor4D | TensorLike,
        strides: number | [number, number],
        pad: "valid" | "same",
        dilation?: number | [number, number],
        dataFormat?: "NHWC" | "NCHW",
    ): T

  • Type Parameters

    Parameters

    • depthwiseFilter: Tensor4D | TensorLike4D
    • pointwiseFilter: Tensor4D | TensorLike
    • strides: number | [number, number]
    • pad: "valid" | "same"
    • Optionaldilation: number | [number, number]
    • OptionaldataFormat: "NHWC" | "NCHW"

    Returns T

sigmoid

  • sigmoid<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

sign

  • sign<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

sin

  • sin<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

sinh

  • sinh<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

slice

  • slice<T extends tf.Tensor>(
        this: T,
        begin: number | number[],
        size?: number | number[],
    ): T

  • Type Parameters

    Parameters

    • this: T
    • begin: number | number[]
    • Optionalsize: number | number[]

    Returns T

softmax

  • softmax<T extends tf.Tensor>(this: T, dim?: number): T

  • Type Parameters

    Parameters

    • this: T
    • Optionaldim: number

    Returns T

softplus

  • softplus<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

spaceToBatchND

  • spaceToBatchND<R extends Rank>(
        blockShape: number[],
        paddings: number[][],
    ): tf.Tensor<R>

  • Type Parameters

    Parameters

    • blockShape: number[]
    • paddings: number[][]

    Returns tf.Tensor<R>

split

  • split<T extends tf.Tensor>(
        numOrSizeSplits: number | number[],
        axis?: number,
    ): T[]

  • Type Parameters

    Parameters

    • numOrSizeSplits: number | number[]
    • Optionalaxis: number

    Returns T[]

sqrt

  • sqrt<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

square

  • square<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

squaredDifference

  • squaredDifference<T extends tf.Tensor>(b: TensorLike | tf.Tensor): T

  • Type Parameters

    Parameters

    Returns T

squeeze

  • squeeze<T extends tf.Tensor>(axis?: number[]): T

  • Type Parameters

    Parameters

    • Optionalaxis: number[]

    Returns T

stack

  • stack<T extends tf.Tensor>(x: tf.Tensor | tf.Tensor[], axis?: number): T

  • Type Parameters

    Parameters

    Returns T

step

  • step<T extends tf.Tensor>(this: T, alpha?: number): T

  • Type Parameters

    Parameters

    • this: T
    • Optionalalpha: number

    Returns T

stridedSlice

  • stridedSlice<T extends tf.Tensor>(
        this: tf.Tensor,
        begin: number[],
        end: number[],
        strides: number[],
        beginMask?: number,
        endMask?: number,
        ellipsisMask?: number,
        newAxisMask?: number,
        shrinkAxisMask?: number,
    ): tf.Tensor

  • Type Parameters

    Parameters

    • this: tf.Tensor
    • begin: number[]
    • end: number[]
    • strides: number[]
    • OptionalbeginMask: number
    • OptionalendMask: number
    • OptionalellipsisMask: number
    • OptionalnewAxisMask: number
    • OptionalshrinkAxisMask: number

    Returns tf.Tensor

sub

sum

  • sum<T extends tf.Tensor>(axis?: number | number[], keepDims?: boolean): T

  • Type Parameters

    Parameters

    • Optionalaxis: number | number[]
    • OptionalkeepDims: boolean

    Returns T

tan

  • tan<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

tanh

  • tanh<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

throwIfDisposed

  • throwIfDisposed(): void

  • Returns void

tile

  • tile<T extends tf.Tensor>(b: number[]): T

  • Type Parameters

    Parameters

    • b: number[]

    Returns T

toBool

  • toBool<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

toFloat

  • toFloat<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

toInt

  • toInt<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

topk

  • topk<T extends tf.Tensor>(
        this: T,
        k?: number,
        sorted?: boolean,
    ): { indices: T; values: T }

  • Type Parameters

    Parameters

    • this: T
    • Optionalk: number
    • Optionalsorted: boolean

    Returns { indices: T; values: T }

toString

  • toString(verbose?: boolean): string

  • Returns a human-readable description of the tensor. Useful for logging.

    Parameters

    • Optionalverbose: boolean

    Returns string

    Doc

transpose

  • transpose<T extends tf.Tensor>(perm?: number[]): T

  • Type Parameters

    Parameters

    • Optionalperm: number[]

    Returns T

unique

  • unique<T extends tf.Tensor>(this: T, axis?: number): { indices: T; values: T }

  • Type Parameters

    Parameters

    • this: T
    • Optionalaxis: number

    Returns { indices: T; values: T }

unsortedSegmentSum

  • unsortedSegmentSum<T extends tf.Tensor>(
        this: T,
        segmentIds: Tensor1D | TensorLike1D,
        numSegments: number,
    ): T

  • Type Parameters

    Parameters

    • this: T
    • segmentIds: Tensor1D | TensorLike1D
    • numSegments: number

    Returns T

unstack

  • unstack<T extends tf.Tensor>(axis?: number): T[]

  • Type Parameters

    Parameters

    • Optionalaxis: number

    Returns T[]

variable

  • variable(
        trainable?: boolean,
        name?: string,
        dtype?: keyof DataTypeMap,
    ): Variable<R>

  • Parameters

    • Optionaltrainable: boolean
    • Optionalname: string
    • Optionaldtype: keyof DataTypeMap

    Returns Variable<R>

where

zerosLike

  • zerosLike<T extends tf.Tensor>(this: T): T

  • Type Parameters

    Parameters

    • this: T

    Returns T

Settings

Member Visibility

On This Page

Constructors
Properties
Accessors
Methods
absacosacoshaddallanyargMaxargMinarrayarraySyncas1Das2Das3Das4Das5DasinasinhasScalarasTypeatanatan2atanhavgPoolbatchNormbatchToSpaceNDbroadcastTobufferbufferSyncbytescastceilclipByValuecloneconcatconv1dconv2dconv2dTransposecoscoshcumprodcumsumdatadataSyncdataToGPUdepthToSpacedepthwiseConv2ddilation2ddisposedivdivNoNandoteluequalerfeuclideanNormexpexpandDimsexpm1fftflattenfloorfloorDivgathergreatergreaterEqualifftirfftisFiniteisInfisNaNleakyRelulesslessEquallocalResponseNormalizationloglog1plogicalAndlogicalNotlogicalOrlogicalXorlogSigmoidlogSoftmaxlogSumExpmatMulmaxmaximummaxPoolmeanminminimummirrorPadmodmulnegnormnotEqualoneHotonesLikepadpoolpowpreluprintprodreciprocalrelurelu6reshapereshapeAsresizeBilinearresizeNearestNeighborreverserfftroundrsqrtseluseparableConv2dsigmoidsignsinsinhslicesoftmaxsoftplusspaceToBatchNDsplitsqrtsquaresquaredDifferencesqueezestackstepstridedSlicesubsumtantanhthrowIfDisposedtiletoBooltoFloattoInttopktoStringtransposeuniqueunsortedSegmentSumunstackvariablewherezerosLike