Hugging Face's logo

Buckets:
hf-doc-build
/
doc-dev

Files
xet
hf-doc-build/doc-dev / transformers.js /pr_1649 /en /api /utils /tensor.md
HuggingFaceDocBuilder's picture
HuggingFaceDocBuilder
preview code
|
download
raw
40.2 kB

utils/tensor

Helper module for Tensor processing.

These functions and classes are only used internally, meaning an end-user shouldn't need to access anything here.

utils/tensor.Tensor

Kind: static class of utils/tensor

new Tensor(...args)

Create a new Tensor or copy an existing Tensor.

  ParamType




...argsArray | Array

tensor.dims : Array

Dimensions of the tensor.

Kind: instance property of Tensor

tensor.type : DataType

Type of the tensor.

Kind: instance property of Tensor

tensor.data : DataArray

The data stored in the tensor.

Kind: instance property of Tensor

tensor.size : number

The number of elements in the tensor.

Kind: instance property of Tensor

tensor.location : string

The location of the tensor data.

Kind: instance property of Tensor

tensor.Symbol.iterator() ⇒ Iterator.<any>

Returns an iterator object for iterating over the tensor data in row-major order. If the tensor has more than one dimension, the iterator will yield subarrays.

Kind: instance method of Tensor
Returns: Iterator.<any> - An iterator object for iterating over the tensor data in row-major order.

tensor._getitem(index) ⇒ Tensor

Index into a Tensor object.

Kind: instance method of Tensor
Returns: Tensor - The data at the specified index. 

  ParamTypeDescription




indexnumberThe index to access.

tensor.indexOf(item) ⇒ number

Kind: instance method of Tensor
Returns: number - The index of the first occurrence of item in the tensor data. 

  ParamTypeDescription




itemnumber | bigintThe item to search for in the tensor

tensor._subarray(index, iterSize, iterDims) ⇒ Tensor

Kind: instance method of Tensor 

  ParamType




indexnumber

iterSizenumber

iterDimsany

tensor.item() ⇒ number | bigint

Returns the value of this tensor as a standard JavaScript Number. This only works for tensors with one element. For other cases, see Tensor.tolist().

Kind: instance method of Tensor
Returns: number | bigint - The value of this tensor as a standard JavaScript Number.
Throws:

  • Error If the tensor has more than one element.

tensor.tolist() ⇒ Array

Convert tensor data to a n-dimensional JS list

Kind: instance method of Tensor

tensor.sigmoid() ⇒ Tensor

Return a new Tensor with the sigmoid function applied to each element.

Kind: instance method of Tensor
Returns: Tensor - The tensor with the sigmoid function applied.

tensor.sigmoid_() ⇒ Tensor

Applies the sigmoid function to the tensor in place.

Kind: instance method of Tensor
Returns: Tensor - Returns this.

tensor.map(callback) ⇒ Tensor

Return a new Tensor with a callback function applied to each element.

Kind: instance method of Tensor
Returns: Tensor - A new Tensor with the callback function applied to each element. 

  ParamTypeDescription




callbackfunctionThe function to apply to each element. It should take three arguments:
                         the current element, its index, and the tensor&#39;s data array.

tensor.map_(callback) ⇒ Tensor

Apply a callback function to each element of the tensor in place.

Kind: instance method of Tensor
Returns: Tensor - Returns this. 

  ParamTypeDescription




callbackfunctionThe function to apply to each element. It should take three arguments:
                         the current element, its index, and the tensor&#39;s data array.

tensor.mul(val) ⇒ Tensor

Return a new Tensor with every element multiplied by a constant.

Kind: instance method of Tensor
Returns: Tensor - The new tensor. 

  ParamTypeDescription




valnumberThe value to multiply by.

tensor.mul_(val) ⇒ Tensor

Multiply the tensor by a constant in place.

Kind: instance method of Tensor
Returns: Tensor - Returns this. 

  ParamTypeDescription




valnumberThe value to multiply by.

tensor.div(val) ⇒ Tensor

Return a new Tensor with every element divided by a constant.

Kind: instance method of Tensor
Returns: Tensor - The new tensor. 

  ParamTypeDescription




valnumberThe value to divide by.

tensor.div_(val) ⇒ Tensor

Divide the tensor by a constant in place.

Kind: instance method of Tensor
Returns: Tensor - Returns this. 

  ParamTypeDescription




valnumberThe value to divide by.

tensor.add(val) ⇒ Tensor

Return a new Tensor with every element added by a constant.

Kind: instance method of Tensor
Returns: Tensor - The new tensor. 

  ParamTypeDescription




valnumberThe value to add by.

tensor.add_(val) ⇒ Tensor

Add the tensor by a constant in place.

Kind: instance method of Tensor
Returns: Tensor - Returns this. 

  ParamTypeDescription




valnumberThe value to add by.

tensor.sub(val) ⇒ Tensor

Return a new Tensor with every element subtracted by a constant.

Kind: instance method of Tensor
Returns: Tensor - The new tensor. 

  ParamTypeDescription




valnumberThe value to subtract by.

tensor.sub_(val) ⇒ Tensor

Subtract the tensor by a constant in place.

Kind: instance method of Tensor
Returns: Tensor - Returns this. 

  ParamTypeDescription




valnumberThe value to subtract by.

tensor.clone() ⇒ Tensor

Creates a deep copy of the current Tensor.

Kind: instance method of Tensor
Returns: Tensor - A new Tensor with the same type, data, and dimensions as the original.

tensor.slice(...slices) ⇒ Tensor

Performs a slice operation on the Tensor along specified dimensions.

Consider a Tensor that has a dimension of [4, 7]:

[ 1,  2,  3,  4,  5,  6,  7]
[ 8,  9, 10, 11, 12, 13, 14]
[15, 16, 17, 18, 19, 20, 21]
[22, 23, 24, 25, 26, 27, 28]

We can slice against the two dims of row and column, for instance in this case we can start at the second element, and return to the second last, like this:

tensor.slice([1, -1], [1, -1]);

which would return:

[  9, 10, 11, 12, 13 ]
[ 16, 17, 18, 19, 20 ]

Kind: instance method of Tensor
Returns: Tensor - A new Tensor containing the selected elements.
Throws:

  • Error If the slice input is invalid.

    ParamTypeDescription

    ...slicesnumber | Array | nullThe slice specifications for each dimension.

If a number is given, then a single element is selected. If an array of two numbers is given, then a range of elements [start, end (exclusive)] is selected. If null is given, then the entire dimension is selected.

tensor.permute(...dims) ⇒ Tensor

Return a permuted version of this Tensor, according to the provided dimensions.

Kind: instance method of Tensor
Returns: Tensor - The permuted tensor. 

  ParamTypeDescription




...dimsnumberDimensions to permute.

tensor.transpose() : Tensor

Kind: instance method of Tensor

tensor.sum([dim], keepdim) ⇒

Returns the sum of each row of the input tensor in the given dimension dim.

Kind: instance method of Tensor
Returns: The summed tensor 

  ParamTypeDefaultDescription




[dim]number | nullThe dimension or dimensions to reduce. If null, all dimensions are reduced.


keepdimbooleanfalseWhether the output tensor has dim retained or not.

tensor.norm([p], [dim], [keepdim]) ⇒ Tensor

Returns the matrix norm or vector norm of a given tensor.

Kind: instance method of Tensor
Returns: Tensor - The norm of the tensor. 

  ParamTypeDefaultDescription




[p]number | string&#x27;fro&#x27;The order of norm


[dim]number | nullSpecifies which dimension of the tensor to calculate the norm across.

If dim is None, the norm will be calculated across all dimensions of input.

[keepdim]booleanfalseWhether the output tensors have dim retained or not.

tensor.normalize_([p], [dim]) ⇒ Tensor

Performs L_p normalization of inputs over specified dimension. Operates in place.

Kind: instance method of Tensor
Returns: Tensor - this for operation chaining. 

  ParamTypeDefaultDescription




[p]number2The exponent value in the norm formulation


[dim]number1The dimension to reduce

tensor.normalize([p], [dim]) ⇒ Tensor

Performs L_p normalization of inputs over specified dimension.

Kind: instance method of Tensor
Returns: Tensor - The normalized tensor. 

  ParamTypeDefaultDescription




[p]number2The exponent value in the norm formulation


[dim]number1The dimension to reduce

tensor.stride() ⇒ Array

Compute and return the stride of this tensor. Stride is the jump necessary to go from one element to the next one in the specified dimension dim.

Kind: instance method of Tensor
Returns: Array - The stride of this tensor.

tensor.squeeze([dim]) ⇒ Tensor

Returns a tensor with all specified dimensions of input of size 1 removed.

NOTE: The returned tensor shares the storage with the input tensor, so changing the contents of one will change the contents of the other. If you would like a copy, use tensor.clone() before squeezing.

Kind: instance method of Tensor
Returns: Tensor - The squeezed tensor 

  ParamTypeDefaultDescription




[dim]number | Array | nullIf given, the input will be squeezed only in the specified dimensions.

tensor.squeeze_()

In-place version of @see Tensor.squeeze

Kind: instance method of Tensor

tensor.unsqueeze(dim) ⇒ Tensor

Returns a new tensor with a dimension of size one inserted at the specified position.

NOTE: The returned tensor shares the same underlying data with this tensor.

Kind: instance method of Tensor
Returns: Tensor - The unsqueezed tensor 

  ParamTypeDescription




dimnumberThe index at which to insert the singleton dimension

tensor.unsqueeze_() : Tensor

In-place version of @see Tensor.unsqueeze

Kind: instance method of Tensor

tensor.flatten_()

In-place version of @see Tensor.flatten

Kind: instance method of Tensor

tensor.flatten(start_dim, end_dim) ⇒ Tensor

Flattens input by reshaping it into a one-dimensional tensor. If start_dim or end_dim are passed, only dimensions starting with start_dim and ending with end_dim are flattened. The order of elements in input is unchanged.

Kind: instance method of Tensor
Returns: Tensor - The flattened tensor. 

  ParamTypeDefaultDescription




start_dimnumber0the first dim to flatten


end_dimnumberthe last dim to flatten

tensor.view(...dims) ⇒ Tensor

Returns a new tensor with the same data as the self tensor but of a different shape.

Kind: instance method of Tensor
Returns: Tensor - The tensor with the same data but different shape 

  ParamTypeDescription




...dimsnumberthe desired size

tensor.gt(val) ⇒ Tensor

Computes input > val element-wise.

Kind: instance method of Tensor
Returns: Tensor - A boolean tensor that is true where input is greater than other and false elsewhere. 

  ParamTypeDescription




valnumberThe value to compare with.

tensor.lt(val) ⇒ Tensor

Computes input Tensor](#module_utils/tensor.Tensor)
Returns: Tensor - A boolean tensor that is true where input is less than other and false elsewhere. 

  ParamTypeDescription




valnumberThe value to compare with.

tensor.clamp_() : Tensor

In-place version of @see Tensor.clamp

Kind: instance method of Tensor

tensor.clamp(min, max) ⇒ Tensor

Clamps all elements in input into the range [ min, max ]

Kind: instance method of Tensor
Returns: Tensor - the output tensor. 

  ParamTypeDescription




minnumberlower-bound of the range to be clamped to


maxnumberupper-bound of the range to be clamped to

tensor.round_()

In-place version of @see Tensor.round

Kind: instance method of Tensor

tensor.round() ⇒ Tensor

Rounds elements of input to the nearest integer.

Kind: instance method of Tensor
Returns: Tensor - the output tensor.

tensor.repeat(...repeats) ⇒ Tensor

Repeats this tensor along the specified dimensions.

Kind: instance method of Tensor
Returns: Tensor - The repeated tensor.
Throws:

  • Error If the number of repeats is less than the number of dimensions.

    ParamTypeDescription

    ...repeatsnumberThe number of times to repeat this tensor along each dimension.

tensor.tile(...dims) ⇒ Tensor

Constructs a tensor by repeating the elements of input. The dims argument specifies the number of repetitions in each dimension.

Kind: instance method of Tensor
Returns: Tensor - The tiled tensor. 

  ParamTypeDescription




...dimsnumberThe number of repetitions per dimension.

tensor.to(type) ⇒ Tensor

Performs Tensor dtype conversion.

Kind: instance method of Tensor
Returns: Tensor - The converted tensor. 

  ParamTypeDescription




typeDataTypeThe desired data type.

utils/tensor.permute(tensor, axes) ⇒ Tensor

Permutes a tensor according to the provided axes.

Kind: static method of utils/tensor
Returns: Tensor - The permuted tensor. 

  ParamTypeDescription




tensoranyThe input tensor to permute.


axesArrayThe axes to permute the tensor along.

utils/tensor.interpolate(input, size, mode, align_corners) ⇒ Tensor

Interpolates an Tensor to the given size.

Kind: static method of utils/tensor
Returns: Tensor - The interpolated tensor. 

  ParamTypeDescription




inputTensorThe input tensor to interpolate. Data must be channel-first (i.e., [c, h, w])


sizeArrayThe output size of the image


modestringThe interpolation mode


align_cornersbooleanWhether to align corners.

utils/tensor.interpolate_4d(input, options) ⇒ Promise.<Tensor>

Down/up samples the input. Inspired by https://pytorch.org/docs/stable/generated/torch.nn.functional.interpolate.html.

Kind: static method of utils/tensor
Returns: Promise.<Tensor> - The interpolated tensor. 

  ParamTypeDefaultDescription




inputTensorthe input tensor


optionsObjectthe options for the interpolation


[options.size]Array | Array | Arrayoutput spatial size.


[options.mode]&quot;nearest&quot; | &quot;bilinear&quot; | &quot;bicubic&quot;&#x27;bilinear&#x27;algorithm used for upsampling

utils/tensor.matmul(a, b) ⇒ Promise.<Tensor>

Matrix product of two tensors. Inspired by https://pytorch.org/docs/stable/generated/torch.matmul.html

Kind: static method of utils/tensor
Returns: Promise.<Tensor> - The matrix product of the two tensors. 

  ParamTypeDescription




aTensorthe first tensor to be multiplied


bTensorthe second tensor to be multiplied

utils/tensor.rfft(x, a) ⇒ Promise.<Tensor>

Computes the one dimensional Fourier transform of real-valued input. Inspired by https://pytorch.org/docs/stable/generated/torch.fft.rfft.html

Kind: static method of utils/tensor
Returns: Promise.<Tensor> - the output tensor. 

  ParamTypeDescription




xTensorthe real input tensor


aTensorThe dimension along which to take the one dimensional real FFT.

utils/tensor.topk(x, [k]) ⇒ Promise.<Array>

Returns the k largest elements of the given input tensor. Inspired by https://pytorch.org/docs/stable/generated/torch.topk.html

Kind: static method of utils/tensor
Returns: Promise.<Array> - the output tuple of (Tensor, LongTensor) of top-k elements and their indices. 

  ParamTypeDescription




xTensorthe input tensor


[k]numberthe k in &quot;top-k&quot;

utils/tensor.slice(data:, starts:, ends:, axes:, [steps]) ⇒ Promise.<Tensor>

Slice a multidimensional float32 tensor.

Kind: static method of utils/tensor
Returns: Promise.<Tensor> - Sliced data tensor. 

  ParamTypeDescription




data:TensorTensor of data to extract slices from


starts:Array1-D array of starting indices of corresponding axis in axes


ends:Array1-D array of ending indices (exclusive) of corresponding axis in axes


axes:Array1-D array of axes that starts and ends apply to


[steps]Array1-D array of slice step of corresponding axis in axes.

utils/tensor.mean_pooling(last_hidden_state, attention_mask) ⇒ Tensor

Perform mean pooling of the last hidden state followed by a normalization step.

Kind: static method of utils/tensor
Returns: Tensor - Returns a new Tensor of shape [batchSize, embedDim]. 

  ParamTypeDescription




last_hidden_stateTensorTensor of shape [batchSize, seqLength, embedDim]


attention_maskTensorTensor of shape [batchSize, seqLength]

utils/tensor.layer_norm(input, normalized_shape, options) ⇒ Tensor

Apply Layer Normalization for last certain number of dimensions.

Kind: static method of utils/tensor
Returns: Tensor - The normalized tensor. 

  ParamTypeDefaultDescription




inputTensorThe input tensor


normalized_shapeArrayinput shape from an expected input of size


optionsObjectThe options for the layer normalization


[options.eps]number1e-5A value added to the denominator for numerical stability.

utils/tensor.cat(tensors, dim) ⇒ Tensor

Concatenates an array of tensors along a specified dimension.

Kind: static method of utils/tensor
Returns: Tensor - The concatenated tensor. 

  ParamTypeDescription




tensorsArrayThe array of tensors to concatenate.


dimnumberThe dimension to concatenate along.

utils/tensor.stack(tensors, dim) ⇒ Tensor

Stack an array of tensors along a specified dimension.

Kind: static method of utils/tensor
Returns: Tensor - The stacked tensor. 

  ParamTypeDescription




tensorsArrayThe array of tensors to stack.


dimnumberThe dimension to stack along.

utils/tensor.std_mean(input, dim, correction, keepdim) ⇒ Array

Calculates the standard deviation and mean over the dimensions specified by dim. dim can be a single dimension or null to reduce over all dimensions.

Kind: static method of utils/tensor
Returns: Array - A tuple of (std, mean) tensors. 

  ParamTypeDescription




inputTensorthe input tenso


dimnumber | nullthe dimension to reduce. If None, all dimensions are reduced.


correctionnumberdifference between the sample size and sample degrees of freedom. Defaults to Bessel&#39;s correction, correction=1.


keepdimbooleanwhether the output tensor has dim retained or not.

utils/tensor.mean(input, dim, keepdim) ⇒ Tensor

Returns the mean value of each row of the input tensor in the given dimension dim.

Kind: static method of utils/tensor
Returns: Tensor - A new tensor with means taken along the specified dimension. 

  ParamTypeDescription




inputTensorthe input tensor.


dimnumber | nullthe dimension to reduce.


keepdimbooleanwhether the output tensor has dim retained or not.

utils/tensor.full(size, fill_value) ⇒ Tensor

Creates a tensor of size size filled with fill_value. The tensor's dtype is inferred from fill_value.

Kind: static method of utils/tensor
Returns: Tensor - The filled tensor. 

  ParamTypeDescription




sizeArrayA sequence of integers defining the shape of the output tensor.


fill_valuenumber | bigint | booleanThe value to fill the output tensor with.

utils/tensor.ones(size) ⇒ Tensor

Returns a tensor filled with the scalar value 1, with the shape defined by the variable argument size.

Kind: static method of utils/tensor
Returns: Tensor - The ones tensor. 

  ParamTypeDescription




sizeArrayA sequence of integers defining the shape of the output tensor.

utils/tensor.ones_like(tensor) ⇒ Tensor

Returns a tensor filled with the scalar value 1, with the same size as input.

Kind: static method of utils/tensor
Returns: Tensor - The ones tensor. 

  ParamTypeDescription




tensorTensorThe size of input will determine size of the output tensor.

utils/tensor.zeros(size) ⇒ Tensor

Returns a tensor filled with the scalar value 0, with the shape defined by the variable argument size.

Kind: static method of utils/tensor
Returns: Tensor - The zeros tensor. 

  ParamTypeDescription




sizeArrayA sequence of integers defining the shape of the output tensor.

utils/tensor.zeros_like(tensor) ⇒ Tensor

Returns a tensor filled with the scalar value 0, with the same size as input.

Kind: static method of utils/tensor
Returns: Tensor - The zeros tensor. 

  ParamTypeDescription




tensorTensorThe size of input will determine size of the output tensor.

utils/tensor.rand(size) ⇒ Tensor

Returns a tensor filled with random numbers from a uniform distribution on the interval [0, 1)

Kind: static method of utils/tensor
Returns: Tensor - The random tensor. 

  ParamTypeDescription




sizeArrayA sequence of integers defining the shape of the output tensor.

utils/tensor.randn(size) ⇒ Tensor

Returns a tensor filled with random numbers from a normal distribution with mean 0 and variance 1 (also called the standard normal distribution).

Kind: static method of utils/tensor
Returns: Tensor - The random tensor. 

  ParamTypeDescription




sizeArrayA sequence of integers defining the shape of the output tensor.

utils/tensor.quantize_embeddings(tensor, precision) ⇒ Tensor

Quantizes the embeddings tensor to binary or unsigned binary precision.

Kind: static method of utils/tensor
Returns: Tensor - The quantized tensor. 

  ParamTypeDescription




tensorTensorThe tensor to quantize.


precision&#x27;binary&#x27; | &#x27;ubinary&#x27;The precision to use for quantization.

utils/tensor~args[0] : ONNXTensor

Kind: inner property of utils/tensor

utils/tensor~reshape(data, dimensions) ⇒ NestArray.<T, DIM>

Reshapes a 1-dimensional array into an n-dimensional array, according to the provided dimensions.

Kind: inner method of utils/tensor
Returns: NestArray.<T, DIM> - The reshaped array. 

  ParamTypeDescription




dataArray | DataArrayThe input array to reshape.


dimensionsDIMThe target shape/dimensions.

Example 

reshape([10                    ], [1      ]); // Type: number[]      Value: [10]
  reshape([1, 2, 3, 4            ], [2, 2   ]); // Type: number[][]    Value: [[1, 2], [3, 4]]
  reshape([1, 2, 3, 4, 5, 6, 7, 8], [2, 2, 2]); // Type: number[][][]  Value: [[[1, 2], [3, 4]], [[5, 6], [7, 8]]]
  reshape([1, 2, 3, 4, 5, 6, 7, 8], [4, 2   ]); // Type: number[][]    Value: [[1, 2], [3, 4], [5, 6], [7, 8]]

reshape~reshapedArray : any

Kind: inner property of reshape

utils/tensor~reduce_helper(callbackfn, input, dim, keepdim, [initialValue]) ⇒ Array

Kind: inner method of utils/tensor
Returns: Array - The reduced tensor data. 

  ParamTypeDefaultDescription




callbackfnfunction

inputTensorthe input tensor.


dimnumberthe dimension to reduce.


keepdimbooleanfalsewhether the output tensor has dim retained or not.


[initialValue]anythe initial value to start the reduction with.

utils/tensor~DataArray : string

Kind: inner typedef of utils/tensor

utils/tensor~NestArray : any

This creates a nested array of a given type and depth (see examples).

Kind: inner typedef of utils/tensor
Example 

NestArray; // string[]

Example 

NestArray; // number[][]

Example 

NestArray; // string[][][] etc.

Xet Storage Details

Size:
40.2 kB
·
Xet hash:
0d5f4e2c95ff2ddb2d53e4184f994272d6c5bdb08b9e82460bd9ace5708020f8

Xet efficiently stores files, intelligently splitting them into unique chunks and accelerating uploads and downloads. More info.