torch.Tensor¶
A torch.Tensor
is a multidimensional matrix containing elements of
a single data type.
Torch defines eight CPU tensor types and eight GPU tensor types:
Data type 
dtype 
CPU tensor 
GPU tensor 

32bit floating point 



64bit floating point 



16bit floating point 



8bit integer (unsigned) 



8bit integer (signed) 



16bit integer (signed) 



32bit integer (signed) 



64bit integer (signed) 



torch.Tensor
is an alias for the default tensor type (torch.FloatTensor
).
A tensor can be constructed from a Python list
or sequence using the
torch.tensor()
constructor:
>>> torch.tensor([[1., 1.], [1., 1.]])
tensor([[ 1.0000, 1.0000],
[ 1.0000, 1.0000]])
>>> torch.tensor(np.array([[1, 2, 3], [4, 5, 6]]))
tensor([[ 1, 2, 3],
[ 4, 5, 6]])
Warning
torch.tensor()
always copies data
. If you have a Tensor
data
and just want to change its requires_grad
flag, use
requires_grad_()
or
detach()
to avoid a copy.
If you have a numpy array and want to avoid a copy, use
torch.as_tensor()
.
A tensor of specific data type can be constructed by passing a
torch.dtype
and/or a torch.device
to a
constructor or tensor creation op:
>>> torch.zeros([2, 4], dtype=torch.int32)
tensor([[ 0, 0, 0, 0],
[ 0, 0, 0, 0]], dtype=torch.int32)
>>> cuda0 = torch.device('cuda:0')
>>> torch.ones([2, 4], dtype=torch.float64, device=cuda0)
tensor([[ 1.0000, 1.0000, 1.0000, 1.0000],
[ 1.0000, 1.0000, 1.0000, 1.0000]], dtype=torch.float64, device='cuda:0')
The contents of a tensor can be accessed and modified using Python’s indexing and slicing notation:
>>> x = torch.tensor([[1, 2, 3], [4, 5, 6]])
>>> print(x[1][2])
tensor(6)
>>> x[0][1] = 8
>>> print(x)
tensor([[ 1, 8, 3],
[ 4, 5, 6]])
Use torch.Tensor.item()
to get a Python number from a tensor containing a
single value:
>>> x = torch.tensor([[1]])
>>> x
tensor([[ 1]])
>>> x.item()
1
>>> x = torch.tensor(2.5)
>>> x
tensor(2.5000)
>>> x.item()
2.5
A tensor can be created with requires_grad=True
so that
torch.autograd
records operations on them for automatic differentiation.
>>> x = torch.tensor([[1., 1.], [1., 1.]], requires_grad=True)
>>> out = x.pow(2).sum()
>>> out.backward()
>>> x.grad
tensor([[ 2.0000, 2.0000],
[ 2.0000, 2.0000]])
Each tensor has an associated torch.Storage
, which holds its data.
The tensor class provides multidimensional, strided
view of a storage and defines numeric operations on it.
Note
For more information on the torch.dtype
, torch.device
, and
torch.layout
attributes of a torch.Tensor
, see
Tensor Attributes.
Note
Methods which mutate a tensor are marked with an underscore suffix.
For example, torch.FloatTensor.abs_()
computes the absolute value
inplace and returns the modified tensor, while torch.FloatTensor.abs()
computes the result in a new tensor.
Note
To change an existing tensor’s torch.device
and/or torch.dtype
, consider using
to()
method on the tensor.

class
torch.
Tensor
¶ There are a few main ways to create a tensor, depending on your use case.
To create a tensor with preexisting data, use
torch.tensor()
.To create a tensor with specific size, use
torch.*
tensor creation ops (see Creation Ops).To create a tensor with the same size (and similar types) as another tensor, use
torch.*_like
tensor creation ops (see Creation Ops).To create a tensor with similar type but different size as another tensor, use
tensor.new_*
creation ops.

new_tensor
(data, dtype=None, device=None, requires_grad=False) → Tensor¶ Returns a new Tensor with
data
as the tensor data. By default, the returned Tensor has the sametorch.dtype
andtorch.device
as this tensor.Warning
new_tensor()
always copiesdata
. If you have a Tensordata
and want to avoid a copy, usetorch.Tensor.requires_grad_()
ortorch.Tensor.detach()
. If you have a numpy array and want to avoid a copy, usetorch.from_numpy()
.Warning
When data is a tensor x,
new_tensor()
reads out ‘the data’ from whatever it is passed, and constructs a leaf variable. Thereforetensor.new_tensor(x)
is equivalent tox.clone().detach()
andtensor.new_tensor(x, requires_grad=True)
is equivalent tox.clone().detach().requires_grad_(True)
. The equivalents usingclone()
anddetach()
are recommended. Parameters
data (array_like) – The returned Tensor copies
data
.dtype (
torch.dtype
, optional) – the desired type of returned tensor. Default: if None, sametorch.dtype
as this tensor.device (
torch.device
, optional) – the desired device of returned tensor. Default: if None, sametorch.device
as this tensor.requires_grad (bool, optional) – If autograd should record operations on the returned tensor. Default:
False
.
Example:
>>> tensor = torch.ones((2,), dtype=torch.int8) >>> data = [[0, 1], [2, 3]] >>> tensor.new_tensor(data) tensor([[ 0, 1], [ 2, 3]], dtype=torch.int8)

new_full
(size, fill_value, dtype=None, device=None, requires_grad=False) → Tensor¶ Returns a Tensor of size
size
filled withfill_value
. By default, the returned Tensor has the sametorch.dtype
andtorch.device
as this tensor. Parameters
fill_value (scalar) – the number to fill the output tensor with.
dtype (
torch.dtype
, optional) – the desired type of returned tensor. Default: if None, sametorch.dtype
as this tensor.device (
torch.device
, optional) – the desired device of returned tensor. Default: if None, sametorch.device
as this tensor.requires_grad (bool, optional) – If autograd should record operations on the returned tensor. Default:
False
.
Example:
>>> tensor = torch.ones((2,), dtype=torch.float64) >>> tensor.new_full((3, 4), 3.141592) tensor([[ 3.1416, 3.1416, 3.1416, 3.1416], [ 3.1416, 3.1416, 3.1416, 3.1416], [ 3.1416, 3.1416, 3.1416, 3.1416]], dtype=torch.float64)

new_empty
(size, dtype=None, device=None, requires_grad=False) → Tensor¶ Returns a Tensor of size
size
filled with uninitialized data. By default, the returned Tensor has the sametorch.dtype
andtorch.device
as this tensor. Parameters
dtype (
torch.dtype
, optional) – the desired type of returned tensor. Default: if None, sametorch.dtype
as this tensor.device (
torch.device
, optional) – the desired device of returned tensor. Default: if None, sametorch.device
as this tensor.requires_grad (bool, optional) – If autograd should record operations on the returned tensor. Default:
False
.
Example:
>>> tensor = torch.ones(()) >>> tensor.new_empty((2, 3)) tensor([[ 5.8182e18, 4.5765e41, 1.0545e+30], [ 3.0949e41, 4.4842e44, 0.0000e+00]])

new_ones
(size, dtype=None, device=None, requires_grad=False) → Tensor¶ Returns a Tensor of size
size
filled with1
. By default, the returned Tensor has the sametorch.dtype
andtorch.device
as this tensor. Parameters
size (int...) – a list, tuple, or
torch.Size
of integers defining the shape of the output tensor.dtype (
torch.dtype
, optional) – the desired type of returned tensor. Default: if None, sametorch.dtype
as this tensor.device (
torch.device
, optional) – the desired device of returned tensor. Default: if None, sametorch.device
as this tensor.requires_grad (bool, optional) – If autograd should record operations on the returned tensor. Default:
False
.
Example:
>>> tensor = torch.tensor((), dtype=torch.int32) >>> tensor.new_ones((2, 3)) tensor([[ 1, 1, 1], [ 1, 1, 1]], dtype=torch.int32)

new_zeros
(size, dtype=None, device=None, requires_grad=False) → Tensor¶ Returns a Tensor of size
size
filled with0
. By default, the returned Tensor has the sametorch.dtype
andtorch.device
as this tensor. Parameters
size (int...) – a list, tuple, or
torch.Size
of integers defining the shape of the output tensor.dtype (
torch.dtype
, optional) – the desired type of returned tensor. Default: if None, sametorch.dtype
as this tensor.device (
torch.device
, optional) – the desired device of returned tensor. Default: if None, sametorch.device
as this tensor.requires_grad (bool, optional) – If autograd should record operations on the returned tensor. Default:
False
.
Example:
>>> tensor = torch.tensor((), dtype=torch.float64) >>> tensor.new_zeros((2, 3)) tensor([[ 0., 0., 0.], [ 0., 0., 0.]], dtype=torch.float64)

is_cuda
¶ Is
True
if the Tensor is stored on the GPU,False
otherwise.

device
¶ Is the
torch.device
where this Tensor is.

grad
¶ This attribute is
None
by default and becomes a Tensor the first time a call tobackward()
computes gradients forself
. The attribute will then contain the gradients computed and future calls tobackward()
will accumulate (add) gradients into it.

abs
() → Tensor¶ See
torch.abs()

acos
() → Tensor¶ See
torch.acos()

add
(value) → Tensor¶ add(value=1, other) > Tensor
See
torch.add()

addbmm
(beta=1, alpha=1, batch1, batch2) → Tensor¶ See
torch.addbmm()

addcdiv
(value=1, tensor1, tensor2) → Tensor¶ See
torch.addcdiv()

addcmul
(value=1, tensor1, tensor2) → Tensor¶ See
torch.addcmul()

addmm
(beta=1, alpha=1, mat1, mat2) → Tensor¶ See
torch.addmm()

addmv
(beta=1, alpha=1, mat, vec) → Tensor¶ See
torch.addmv()

addr
(beta=1, alpha=1, vec1, vec2) → Tensor¶ See
torch.addr()

allclose
(other, rtol=1e05, atol=1e08, equal_nan=False) → Tensor¶ See
torch.allclose()

apply_
(callable) → Tensor¶ Applies the function
callable
to each element in the tensor, replacing each element with the value returned bycallable
.Note
This function only works with CPU tensors and should not be used in code sections that require high performance.

argmax
(dim=None, keepdim=False) → LongTensor¶ See
torch.argmax()

argmin
(dim=None, keepdim=False) → LongTensor¶ See
torch.argmin()

argsort
(dim=1, descending=False) → LongTensor¶ See :func: torch.argsort

asin
() → Tensor¶ See
torch.asin()

atan
() → Tensor¶ See
torch.atan()

atan2
(other) → Tensor¶ See
torch.atan2()

backward
(gradient=None, retain_graph=None, create_graph=False)[source]¶ Computes the gradient of current tensor w.r.t. graph leaves.
The graph is differentiated using the chain rule. If the tensor is nonscalar (i.e. its data has more than one element) and requires gradient, the function additionally requires specifying
gradient
. It should be a tensor of matching type and location, that contains the gradient of the differentiated function w.r.t.self
.This function accumulates gradients in the leaves  you might need to zero them before calling it.
 Parameters
gradient (Tensor or None) – Gradient w.r.t. the tensor. If it is a tensor, it will be automatically converted to a Tensor that does not require grad unless
create_graph
is True. None values can be specified for scalar Tensors or ones that don’t require grad. If a None value would be acceptable then this argument is optional.retain_graph (bool, optional) – If
False
, the graph used to compute the grads will be freed. Note that in nearly all cases setting this option to True is not needed and often can be worked around in a much more efficient way. Defaults to the value ofcreate_graph
.create_graph (bool, optional) – If
True
, graph of the derivative will be constructed, allowing to compute higher order derivative products. Defaults toFalse
.

baddbmm
(beta=1, alpha=1, batch1, batch2) → Tensor¶ See
torch.baddbmm()

bernoulli
(*, generator=None) → Tensor¶ Returns a result tensor where each \(\texttt{result[i]}\) is independently sampled from \(\text{Bernoulli}(\texttt{self[i]})\).
self
must have floating pointdtype
, and the result will have the samedtype
.

bernoulli_
()¶ 
bernoulli_
(p=0.5, *, generator=None) → Tensor Fills each location of
self
with an independent sample from \(\text{Bernoulli}(\texttt{p})\).self
can have integraldtype
.

bernoulli_
(p_tensor, *, generator=None) → Tensor p_tensor
should be a tensor containing probabilities to be used for drawing the binary random number.The \(\text{i}^{th}\) element of
self
tensor will be set to a value sampled from \(\text{Bernoulli}(\texttt{p\_tensor[i]})\).self
can have integraldtype
, butp_tensor
must have floating pointdtype
.
See also
bernoulli()
andtorch.bernoulli()


bincount
(weights=None, minlength=0) → Tensor¶ See
torch.bincount()

bmm
(batch2) → Tensor¶ See
torch.bmm()

btrifact
(pivot=True)[source]¶ See
torch.lu()

btrifact_with_info
(pivot=True)[source]¶ See
torch.lu()

btrisolve
(LU_data, LU_pivots)[source]¶ See
torch.lu_solve()

cauchy_
(median=0, sigma=1, *, generator=None) → Tensor¶ Fills the tensor with numbers drawn from the Cauchy distribution:
\[f(x) = \dfrac{1}{\pi} \dfrac{\sigma}{(x  \text{median})^2 + \sigma^2}\]

ceil
() → Tensor¶ See
torch.ceil()

cholesky
(upper=False) → Tensor¶ See
torch.cholesky()

cholesky_inverse
(upper=False) → Tensor¶

cholesky_solve
(input2, upper=False) → Tensor¶

chunk
(chunks, dim=0) → List of Tensors¶ See
torch.chunk()

clamp
(min, max) → Tensor¶ See
torch.clamp()

clone
() → Tensor¶ Returns a copy of the
self
tensor. The copy has the same size and data type asself
.Note
Unlike copy_(), this function is recorded in the computation graph. Gradients propagating to the cloned tensor will propagate to the original tensor.

contiguous
() → Tensor¶ Returns a contiguous tensor containing the same data as
self
tensor. Ifself
tensor is contiguous, this function returns theself
tensor.

copy_
(src, non_blocking=False) → Tensor¶ Copies the elements from
src
intoself
tensor and returnsself
.The
src
tensor must be broadcastable with theself
tensor. It may be of a different data type or reside on a different device.

cos
() → Tensor¶ See
torch.cos()

cosh
() → Tensor¶ See
torch.cosh()

cpu
() → Tensor¶ Returns a copy of this object in CPU memory.
If this object is already in CPU memory and on the correct device, then no copy is performed and the original object is returned.

cross
(other, dim=1) → Tensor¶ See
torch.cross()

cuda
(device=None, non_blocking=False) → Tensor¶ Returns a copy of this object in CUDA memory.
If this object is already in CUDA memory and on the correct device, then no copy is performed and the original object is returned.
 Parameters
device (
torch.device
) – The destination GPU device. Defaults to the current CUDA device.non_blocking (bool) – If
True
and the source is in pinned memory, the copy will be asynchronous with respect to the host. Otherwise, the argument has no effect. Default:False
.

cumprod
(dim, dtype=None) → Tensor¶ See
torch.cumprod()

cumsum
(dim, dtype=None) → Tensor¶ See
torch.cumsum()

data_ptr
() → int¶ Returns the address of the first element of
self
tensor.

dequantize
() → Tensor¶ Given a quantized Tensor, dequantize it and return the dequantized float Tensor.

det
() → Tensor¶ See
torch.det()

dense_dim
() → int¶ If
self
is a sparse COO tensor (i.e., withtorch.sparse_coo
layout), this returns a the number of dense dimensions. Otherwise, this throws an error.See also
Tensor.sparse_dim()
.

detach
()¶ Returns a new Tensor, detached from the current graph.
The result will never require gradient.
Note
Returned Tensor shares the same storage with the original one. Inplace modifications on either of them will be seen, and may trigger errors in correctness checks. IMPORTANT NOTE: Previously, inplace size / stride / storage changes (such as resize_ / resize_as_ / set_ / transpose_) to the returned tensor also update the original tensor. Now, these inplace changes will not update the original tensor anymore, and will instead trigger an error. For sparse tensors: Inplace indices / values changes (such as zero_ / copy_ / add_) to the returned tensor will not update the original tensor anymore, and will instead trigger an error.

detach_
()¶ Detaches the Tensor from the graph that created it, making it a leaf. Views cannot be detached inplace.

diag
(diagonal=0) → Tensor¶ See
torch.diag()

diag_embed
(offset=0, dim1=2, dim2=1) → Tensor¶

diagflat
(diagonal=0) → Tensor¶ See
torch.diagflat()

diagonal
(offset=0, dim1=0, dim2=1) → Tensor¶ See
torch.diagonal()

digamma
() → Tensor¶ See
torch.digamma()

dim
() → int¶ Returns the number of dimensions of
self
tensor.

dist
(other, p=2) → Tensor¶ See
torch.dist()

div
(value) → Tensor¶ See
torch.div()

dot
(tensor2) → Tensor¶ See
torch.dot()

eig
(eigenvectors=False) > (Tensor, Tensor)¶ See
torch.eig()

element_size
() → int¶ Returns the size in bytes of an individual element.
Example:
>>> torch.tensor([]).element_size() 4 >>> torch.tensor([], dtype=torch.uint8).element_size() 1

eq
(other) → Tensor¶ See
torch.eq()

equal
(other) → bool¶ See
torch.equal()

erf
() → Tensor¶ See
torch.erf()

erfc
() → Tensor¶ See
torch.erfc()

erfinv
() → Tensor¶ See
torch.erfinv()

exp
() → Tensor¶ See
torch.exp()

expm1
() → Tensor¶ See
torch.expm1()

expand
(*sizes) → Tensor¶ Returns a new view of the
self
tensor with singleton dimensions expanded to a larger size.Passing 1 as the size for a dimension means not changing the size of that dimension.
Tensor can be also expanded to a larger number of dimensions, and the new ones will be appended at the front. For the new dimensions, the size cannot be set to 1.
Expanding a tensor does not allocate new memory, but only creates a new view on the existing tensor where a dimension of size one is expanded to a larger size by setting the
stride
to 0. Any dimension of size 1 can be expanded to an arbitrary value without allocating new memory. Parameters
*sizes (torch.Size or int...) – the desired expanded size
Warning
More than one element of an expanded tensor may refer to a single memory location. As a result, inplace operations (especially ones that are vectorized) may result in incorrect behavior. If you need to write to the tensors, please clone them first.
Example:
>>> x = torch.tensor([[1], [2], [3]]) >>> x.size() torch.Size([3, 1]) >>> x.expand(3, 4) tensor([[ 1, 1, 1, 1], [ 2, 2, 2, 2], [ 3, 3, 3, 3]]) >>> x.expand(1, 4) # 1 means not changing the size of that dimension tensor([[ 1, 1, 1, 1], [ 2, 2, 2, 2], [ 3, 3, 3, 3]])

expand_as
(other) → Tensor¶ Expand this tensor to the same size as
other
.self.expand_as(other)
is equivalent toself.expand(other.size())
.Please see
expand()
for more information aboutexpand
. Parameters
other (
torch.Tensor
) – The result tensor has the same size asother
.

exponential_
(lambd=1, *, generator=None) → Tensor¶ Fills
self
tensor with elements drawn from the exponential distribution:\[f(x) = \lambda e^{\lambda x}\]

fft
(signal_ndim, normalized=False) → Tensor¶ See
torch.fft()

fill_
(value) → Tensor¶ Fills
self
tensor with the specified value.

flatten
(input, start_dim=0, end_dim=1) → Tensor¶ see
torch.flatten()

flip
(dims) → Tensor¶ See
torch.flip()

floor
() → Tensor¶ See
torch.floor()

fmod
(divisor) → Tensor¶ See
torch.fmod()

frac
() → Tensor¶ See
torch.frac()

gather
(dim, index) → Tensor¶ See
torch.gather()

ge
(other) → Tensor¶ See
torch.ge()

gels
(A) → Tensor¶ See
torch.gels()

geometric_
(p, *, generator=None) → Tensor¶ Fills
self
tensor with elements drawn from the geometric distribution:\[f(X=k) = (1  p)^{k  1} p\]

geqrf
() > (Tensor, Tensor)¶ See
torch.geqrf()

ger
(vec2) → Tensor¶ See
torch.ger()

gesv
(A)[source]¶ See
torch.solve()

get_device
() > Device ordinal (Integer)¶ For CUDA tensors, this function returns the device ordinal of the GPU on which the tensor resides. For CPU tensors, an error is thrown.
Example:
>>> x = torch.randn(3, 4, 5, device='cuda:0') >>> x.get_device() 0 >>> x.cpu().get_device() # RuntimeError: get_device is not implemented for type torch.FloatTensor

gt
(other) → Tensor¶ See
torch.gt()

hardshrink
(lambd=0.5) → Tensor¶

histc
(bins=100, min=0, max=0) → Tensor¶ See
torch.histc()

ifft
(signal_ndim, normalized=False) → Tensor¶ See
torch.ifft()

index_add_
(dim, index, tensor) → Tensor¶ Accumulate the elements of
tensor
into theself
tensor by adding to the indices in the order given inindex
. For example, ifdim == 0
andindex[i] == j
, then thei
th row oftensor
is added to thej
th row ofself
.The
dim
th dimension oftensor
must have the same size as the length ofindex
(which must be a vector), and all other dimensions must matchself
, or an error will be raised.Note
When using the CUDA backend, this operation may induce nondeterministic behaviour that is not easily switched off. Please see the notes on Reproducibility for background.
 Parameters
Example:
>>> x = torch.ones(5, 3) >>> t = torch.tensor([[1, 2, 3], [4, 5, 6], [7, 8, 9]], dtype=torch.float) >>> index = torch.tensor([0, 4, 2]) >>> x.index_add_(0, index, t) tensor([[ 2., 3., 4.], [ 1., 1., 1.], [ 8., 9., 10.], [ 1., 1., 1.], [ 5., 6., 7.]])

index_add
(dim, index, tensor) → Tensor¶ Outofplace version of
torch.Tensor.index_add_()

index_copy_
(dim, index, tensor) → Tensor¶ Copies the elements of
tensor
into theself
tensor by selecting the indices in the order given inindex
. For example, ifdim == 0
andindex[i] == j
, then thei
th row oftensor
is copied to thej
th row ofself
.The
dim
th dimension oftensor
must have the same size as the length ofindex
(which must be a vector), and all other dimensions must matchself
, or an error will be raised. Parameters
Example:
>>> x = torch.zeros(5, 3) >>> t = torch.tensor([[1, 2, 3], [4, 5, 6], [7, 8, 9]], dtype=torch.float) >>> index = torch.tensor([0, 4, 2]) >>> x.index_copy_(0, index, t) tensor([[ 1., 2., 3.], [ 0., 0., 0.], [ 7., 8., 9.], [ 0., 0., 0.], [ 4., 5., 6.]])

index_copy
(dim, index, tensor) → Tensor¶ Outofplace version of
torch.Tensor.index_copy_()

index_fill_
(dim, index, val) → Tensor¶ Fills the elements of the
self
tensor with valueval
by selecting the indices in the order given inindex
. Parameters
 Example::
>>> x = torch.tensor([[1, 2, 3], [4, 5, 6], [7, 8, 9]], dtype=torch.float) >>> index = torch.tensor([0, 2]) >>> x.index_fill_(1, index, 1) tensor([[1., 2., 1.], [1., 5., 1.], [1., 8., 1.]])

index_fill
(dim, index, value) → Tensor¶ Outofplace version of
torch.Tensor.index_fill_()

index_put_
(indices, value, accumulate=False) → Tensor¶ Puts values from the tensor
value
into the tensorself
using the indices specified inindices
(which is a tuple of Tensors). The expressiontensor.index_put_(indices, value)
is equivalent totensor[indices] = value
. Returnsself
.If
accumulate
isTrue
, the elements intensor
are added toself
. If accumulate isFalse
, the behavior is undefined if indices contain duplicate elements.

index_put
(indices, value, accumulate=False) → Tensor¶ Outplace version of
index_put_()

index_select
(dim, index) → Tensor¶

indices
() → Tensor¶ If
self
is a sparse COO tensor (i.e., withtorch.sparse_coo
layout), this returns a view of the contained indices tensor. Otherwise, this throws an error.See also
Tensor.values()
.Note
This method can only be called on a coalesced sparse tensor. See
Tensor.coalesce()
for details.

int_repr
() → Tensor¶ Given a quantized Tensor,
self.int_repr()
returns a CPU Tensor with uint8_t as data type that stores the underlying uint8_t values of the given Tensor.

inverse
() → Tensor¶ See
torch.inverse()

irfft
(signal_ndim, normalized=False, onesided=True, signal_sizes=None) → Tensor¶ See
torch.irfft()

is_contiguous
() → bool¶ Returns True if
self
tensor is contiguous in memory in C order.

is_floating_point
() → bool¶ Returns True if the data type of
self
is a floating point data type.

is_leaf
()¶ All Tensors that have
requires_grad
which isFalse
will be leaf Tensors by convention.For Tensors that have
requires_grad
which isTrue
, they will be leaf Tensors if they were created by the user. This means that they are not the result of an operation and sograd_fn
is None.Only leaf Tensors will have their
grad
populated during a call tobackward()
. To getgrad
populated for nonleaf Tensors, you can useretain_grad()
.Example:
>>> a = torch.rand(10, requires_grad=True) >>> a.is_leaf True >>> b = torch.rand(10, requires_grad=True).cuda() >>> b.is_leaf False # b was created by the operation that cast a cpu Tensor into a cuda Tensor >>> c = torch.rand(10, requires_grad=True) + 2 >>> c.is_leaf False # c was created by the addition operation >>> d = torch.rand(10).cuda() >>> d.is_leaf True # d does not require gradients and so has no operation creating it (that is tracked by the autograd engine) >>> e = torch.rand(10).cuda().requires_grad_() >>> e.is_leaf True # e requires gradients and has no operations creating it >>> f = torch.rand(10, requires_grad=True, device="cuda") >>> f.is_leaf True # f requires grad, has no operation creating it

is_set_to
(tensor) → bool¶ Returns True if this object refers to the same
THTensor
object from the Torch C API as the given tensor.
Checks if tensor is in shared memory.
This is always
True
for CUDA tensors.

is_signed
() → bool¶ Returns True if the data type of
self
is a signed data type.

is_sparse
()¶

item
() → number¶ Returns the value of this tensor as a standard Python number. This only works for tensors with one element. For other cases, see
tolist()
.This operation is not differentiable.
Example:
>>> x = torch.tensor([1.0]) >>> x.item() 1.0

kthvalue
(k, dim=None, keepdim=False) > (Tensor, LongTensor)¶ See
torch.kthvalue()

le
(other) → Tensor¶ See
torch.le()

lerp
(end, weight) → Tensor¶ See
torch.lerp()

log
() → Tensor¶ See
torch.log()

logdet
() → Tensor¶ See
torch.logdet()

log10
() → Tensor¶ See
torch.log10()

log1p
() → Tensor¶ See
torch.log1p()

log2
() → Tensor¶ See
torch.log2()

log_normal_
(mean=1, std=2, *, generator=None)¶ Fills
self
tensor with numbers samples from the lognormal distribution parameterized by the given mean \(\mu\) and standard deviation \(\sigma\). Note thatmean
andstd
are the mean and standard deviation of the underlying normal distribution, and not of the returned distribution:\[f(x) = \dfrac{1}{x \sigma \sqrt{2\pi}}\ e^{\frac{(\ln x  \mu)^2}{2\sigma^2}}\]

logsumexp
(dim, keepdim=False) → Tensor¶

lt
(other) → Tensor¶ See
torch.lt()

lu
(pivot=True, get_infos=False)[source]¶ See
torch.lu()

lu_solve
(LU_data, LU_pivots) → Tensor¶ See
torch.lu_solve()

map_
(tensor, callable)¶ Applies
callable
for each element inself
tensor and the giventensor
and stores the results inself
tensor.self
tensor and the giventensor
must be broadcastable.The
callable
should have the signature:def callable(a, b) > number

masked_scatter_
(mask, source)¶ Copies elements from
source
intoself
tensor at positions where themask
is one. The shape ofmask
must be broadcastable with the shape of the underlying tensor. Thesource
should have at least as many elements as the number of ones inmask
 Parameters
mask (ByteTensor) – the binary mask
source (Tensor) – the tensor to copy from
Note
The
mask
operates on theself
tensor, not on the givensource
tensor.

masked_scatter
(mask, tensor) → Tensor¶ Outofplace version of
torch.Tensor.masked_scatter_()

masked_fill_
(mask, value)¶ Fills elements of
self
tensor withvalue
wheremask
is one. The shape ofmask
must be broadcastable with the shape of the underlying tensor. Parameters
mask (ByteTensor) – the binary mask
value (float) – the value to fill in with

masked_fill
(mask, value) → Tensor¶ Outofplace version of
torch.Tensor.masked_fill_()

masked_select
(mask) → Tensor¶

matmul
(tensor2) → Tensor¶ See
torch.matmul()

matrix_power
(n) → Tensor¶

max
(dim=None, keepdim=False) > Tensor or (Tensor, Tensor)¶ See
torch.max()

mean
(dim=None, keepdim=False) > Tensor or (Tensor, Tensor)¶ See
torch.mean()

median
(dim=None, keepdim=False) > (Tensor, LongTensor)¶ See
torch.median()

min
(dim=None, keepdim=False) > Tensor or (Tensor, Tensor)¶ See
torch.min()

mm
(mat2) → Tensor¶ See
torch.mm()

mode
(dim=None, keepdim=False) > (Tensor, LongTensor)¶ See
torch.mode()

mul
(value) → Tensor¶ See
torch.mul()

multinomial
(num_samples, replacement=False, *, generator=None) → Tensor¶

mv
(vec) → Tensor¶ See
torch.mv()

mvlgamma
(p) → Tensor¶ See
torch.mvlgamma()

mvlgamma_
(p) → Tensor¶ Inplace version of
mvlgamma()

narrow
(dimension, start, length) → Tensor¶ See
torch.narrow()
Example:
>>> x = torch.tensor([[1, 2, 3], [4, 5, 6], [7, 8, 9]]) >>> x.narrow(0, 0, 2) tensor([[ 1, 2, 3], [ 4, 5, 6]]) >>> x.narrow(1, 1, 2) tensor([[ 2, 3], [ 5, 6], [ 8, 9]])

narrow_copy
(dimension, start, length) → Tensor¶ Same as
Tensor.narrow()
except returning a copy rather than shared storage. This is primarily for sparse tensors, which do not have a sharedstorage narrow method. Calling`narrow_copy
with`dimemsion > self.sparse_dim()`
will return a copy with the relevant dense dimension narrowed, and`self.shape`
updated accordingly.

ne
(other) → Tensor¶ See
torch.ne()

neg
() → Tensor¶ See
torch.neg()

nonzero
() → LongTensor¶ See
torch.nonzero()

norm
(p='fro', dim=None, keepdim=False, dtype=None)[source]¶ See
torch.norm()

normal_
(mean=0, std=1, *, generator=None) → Tensor¶ Fills
self
tensor with elements samples from the normal distribution parameterized bymean
andstd
.

numel
() → int¶ See
torch.numel()

numpy
() → numpy.ndarray¶ Returns
self
tensor as a NumPyndarray
. This tensor and the returnedndarray
share the same underlying storage. Changes toself
tensor will be reflected in thendarray
and vice versa.

orgqr
(input2) → Tensor¶ See
torch.orgqr()

ormqr
(input2, input3, left=True, transpose=False) → Tensor¶ See
torch.ormqr()

permute
(*dims) → Tensor¶ Permute the dimensions of this tensor.
 Parameters
*dims (int...) – The desired ordering of dimensions
Example
>>> x = torch.randn(2, 3, 5) >>> x.size() torch.Size([2, 3, 5]) >>> x.permute(2, 0, 1).size() torch.Size([5, 2, 3])

pin_memory
() → Tensor¶ Copies the tensor to pinned memory, if it’s not already pinned.

pinverse
() → Tensor¶ See
torch.pinverse()

potrf
(upper=True)[source]¶ See
torch.cholesky()

pow
(exponent) → Tensor¶ See
torch.pow()

prod
(dim=None, keepdim=False, dtype=None) → Tensor¶ See
torch.prod()

pstrf
(upper=True)[source]¶ See
torch.pstrf()

put_
(indices, tensor, accumulate=False) → Tensor¶ Copies the elements from
tensor
into the positions specified by indices. For the purpose of indexing, theself
tensor is treated as if it were a 1D tensor.If
accumulate
isTrue
, the elements intensor
are added toself
. If accumulate isFalse
, the behavior is undefined if indices contain duplicate elements. Parameters
Example:
>>> src = torch.tensor([[4, 3, 5], [6, 7, 8]]) >>> src.put_(torch.tensor([1, 3]), torch.tensor([9, 10])) tensor([[ 4, 9, 5], [ 10, 7, 8]])

qr
() > (Tensor, Tensor)¶ See
torch.qr()

quantize_linear
(scale, zero_point) → Tensor¶ Quantize a float Tensor using affine quantization scheme with given scale and zero_point. returns the quantized Tensor.

q_scale
() → float¶ Given a Tensor quantized by linear(affine) quantization, returns the scale of the underlying quantizer().

q_zero_point
() → int¶ Given a Tensor quantized by linear(affine) quantization, returns the zero_point of the underlying quantizer().

random_
(from=0, to=None, *, generator=None) → Tensor¶ Fills
self
tensor with numbers sampled from the discrete uniform distribution over[from, to  1]
. If not specified, the values are usually only bounded byself
tensor’s data type. However, for floating point types, if unspecified, range will be[0, 2^mantissa]
to ensure that every value is representable. For example, torch.tensor(1, dtype=torch.double).random_() will be uniform in[0, 2^53]
.

reciprocal
() → Tensor¶

reciprocal_
() → Tensor¶ Inplace version of
reciprocal()

register_hook
(hook)[source]¶ Registers a backward hook.
The hook will be called every time a gradient with respect to the Tensor is computed. The hook should have the following signature:
hook(grad) > Tensor or None
The hook should not modify its argument, but it can optionally return a new gradient which will be used in place of
grad
.This function returns a handle with a method
handle.remove()
that removes the hook from the module.Example:
>>> v = torch.tensor([0., 0., 0.], requires_grad=True) >>> h = v.register_hook(lambda grad: grad * 2) # double the gradient >>> v.backward(torch.tensor([1., 2., 3.])) >>> v.grad 2 4 6 [torch.FloatTensor of size (3,)] >>> h.remove() # removes the hook

remainder
(divisor) → Tensor¶

remainder_
(divisor) → Tensor¶ Inplace version of
remainder()

renorm
(p, dim, maxnorm) → Tensor¶ See
torch.renorm()

repeat
(*sizes) → Tensor¶ Repeats this tensor along the specified dimensions.
Unlike
expand()
, this function copies the tensor’s data.Warning
torch.repeat()
behaves differently from numpy.repeat, but is more similar to numpy.tile. For the operator similar to numpy.repeat, seetorch.repeat_interleave()
. Parameters
sizes (torch.Size or int...) – The number of times to repeat this tensor along each dimension
Example:
>>> x = torch.tensor([1, 2, 3]) >>> x.repeat(4, 2) tensor([[ 1, 2, 3, 1, 2, 3], [ 1, 2, 3, 1, 2, 3], [ 1, 2, 3, 1, 2, 3], [ 1, 2, 3, 1, 2, 3]]) >>> x.repeat(4, 2, 1).size() torch.Size([4, 2, 3])

repeat_interleave
(repeats, dim=None) → Tensor¶

requires_grad
()¶ Is
True
if gradients need to be computed for this Tensor,False
otherwise.

requires_grad_
(requires_grad=True) → Tensor¶ Change if autograd should record operations on this tensor: sets this tensor’s
requires_grad
attribute inplace. Returns this tensor.require_grad_()
’s main use case is to tell autograd to begin recording operations on a Tensortensor
. Iftensor
hasrequires_grad=False
(because it was obtained through a DataLoader, or required preprocessing or initialization),tensor.requires_grad_()
makes it so that autograd will begin to record operations ontensor
. Parameters
requires_grad (bool) – If autograd should record operations on this tensor. Default:
True
.
Example:
>>> # Let's say we want to preprocess some saved weights and use >>> # the result as new weights. >>> saved_weights = [0.1, 0.2, 0.3, 0.25] >>> loaded_weights = torch.tensor(saved_weights) >>> weights = preprocess(loaded_weights) # some function >>> weights tensor([0.5503, 0.4926, 2.1158, 0.8303]) >>> # Now, start to record operations done to weights >>> weights.requires_grad_() >>> out = weights.pow(2).sum() >>> out.backward() >>> weights.grad tensor([1.1007, 0.9853, 4.2316, 1.6606])

reshape
(*shape) → Tensor¶ Returns a tensor with the same data and number of elements as
self
but with the specified shape. This method returns a view ifshape
is compatible with the current shape. Seetorch.Tensor.view()
on when it is possible to return a view.See
torch.reshape()
 Parameters
shape (tuple of python:ints or int...) – the desired shape

reshape_as
(other) → Tensor¶ Returns this tensor as the same shape as
other
.self.reshape_as(other)
is equivalent toself.reshape(other.sizes())
. This method returns a view ifother.sizes()
is compatible with the current shape. Seetorch.Tensor.view()
on when it is possible to return a view.Please see
reshape()
for more information aboutreshape
. Parameters
other (
torch.Tensor
) – The result tensor has the same shape asother
.

resize_
(*sizes) → Tensor¶ Resizes
self
tensor to the specified size. If the number of elements is larger than the current storage size, then the underlying storage is resized to fit the new number of elements. If the number of elements is smaller, the underlying storage is not changed. Existing elements are preserved but any new memory is uninitialized.Warning
This is a lowlevel method. The storage is reinterpreted as Ccontiguous, ignoring the current strides (unless the target size equals the current size, in which case the tensor is left unchanged). For most purposes, you will instead want to use
view()
, which checks for contiguity, orreshape()
, which copies data if needed. To change the size inplace with custom strides, seeset_()
. Parameters
sizes (torch.Size or int...) – the desired size
Example:
>>> x = torch.tensor([[1, 2], [3, 4], [5, 6]]) >>> x.resize_(2, 2) tensor([[ 1, 2], [ 3, 4]])

resize_as_
(tensor) → Tensor¶ Resizes the
self
tensor to be the same size as the specifiedtensor
. This is equivalent toself.resize_(tensor.size())
.

rfft
(signal_ndim, normalized=False, onesided=True) → Tensor¶ See
torch.rfft()

roll
(shifts, dims) → Tensor¶ See
torch.roll()

rot90
(k, dims) → Tensor¶ See
torch.rot90()

round
() → Tensor¶ See
torch.round()

rsqrt
() → Tensor¶ See
torch.rsqrt()

scatter
(dim, index, source) → Tensor¶ Outofplace version of
torch.Tensor.scatter_()

scatter_
(dim, index, src) → Tensor¶ Writes all values from the tensor
src
intoself
at the indices specified in theindex
tensor. For each value insrc
, its output index is specified by its index insrc
fordimension != dim
and by the corresponding value inindex
fordimension = dim
.For a 3D tensor,
self
is updated as:self[index[i][j][k]][j][k] = src[i][j][k] # if dim == 0 self[i][index[i][j][k]][k] = src[i][j][k] # if dim == 1 self[i][j][index[i][j][k]] = src[i][j][k] # if dim == 2
This is the reverse operation of the manner described in
gather()
.self
,index
andsrc
(if it is a Tensor) should have same number of dimensions. It is also required thatindex.size(d) <= src.size(d)
for all dimensionsd
, and thatindex.size(d) <= self.size(d)
for all dimensionsd != dim
.Moreover, as for
gather()
, the values ofindex
must be between0
andself.size(dim)  1
inclusive, and all values in a row along the specified dimensiondim
must be unique. Parameters
dim (int) – the axis along which to index
index (LongTensor) – the indices of elements to scatter, can be either empty or the same size of src. When empty, the operation returns identity
src (Tensor) – the source element(s) to scatter, incase value is not specified
value (float) – the source element(s) to scatter, incase src is not specified
Example:
>>> x = torch.rand(2, 5) >>> x tensor([[ 0.3992, 0.2908, 0.9044, 0.4850, 0.6004], [ 0.5735, 0.9006, 0.6797, 0.4152, 0.1732]]) >>> torch.zeros(3, 5).scatter_(0, torch.tensor([[0, 1, 2, 0, 0], [2, 0, 0, 1, 2]]), x) tensor([[ 0.3992, 0.9006, 0.6797, 0.4850, 0.6004], [ 0.0000, 0.2908, 0.0000, 0.4152, 0.0000], [ 0.5735, 0.0000, 0.9044, 0.0000, 0.1732]]) >>> z = torch.zeros(2, 4).scatter_(1, torch.tensor([[2], [3]]), 1.23) >>> z tensor([[ 0.0000, 0.0000, 1.2300, 0.0000], [ 0.0000, 0.0000, 0.0000, 1.2300]])

scatter_add_
(dim, index, other) → Tensor¶ Adds all values from the tensor
other
intoself
at the indices specified in theindex
tensor in a similar fashion asscatter_()
. For each value inother
, it is added to an index inself
which is specified by its index inother
fordimension != dim
and by the corresponding value inindex
fordimension = dim
.For a 3D tensor,
self
is updated as:self[index[i][j][k]][j][k] += other[i][j][k] # if dim == 0 self[i][index[i][j][k]][k] += other[i][j][k] # if dim == 1 self[i][j][index[i][j][k]] += other[i][j][k] # if dim == 2
self
,index
andother
should have same number of dimensions. It is also required thatindex.size(d) <= other.size(d)
for all dimensionsd
, and thatindex.size(d) <= self.size(d)
for all dimensionsd != dim
.Moreover, as for
gather()
, the values ofindex
must be between0
andself.size(dim)  1
inclusive, and all values in a row along the specified dimensiondim
must be unique.Note
When using the CUDA backend, this operation may induce nondeterministic behaviour that is not easily switched off. Please see the notes on Reproducibility for background.
 Parameters
Example:
>>> x = torch.rand(2, 5) >>> x tensor([[0.7404, 0.0427, 0.6480, 0.3806, 0.8328], [0.7953, 0.2009, 0.9154, 0.6782, 0.9620]]) >>> torch.ones(3, 5).scatter_add_(0, torch.tensor([[0, 1, 2, 0, 0], [2, 0, 0, 1, 2]]), x) tensor([[1.7404, 1.2009, 1.9154, 1.3806, 1.8328], [1.0000, 1.0427, 1.0000, 1.6782, 1.0000], [1.7953, 1.0000, 1.6480, 1.0000, 1.9620]])

scatter_add
(dim, index, source) → Tensor¶ Outofplace version of
torch.Tensor.scatter_add_()

select
(dim, index) → Tensor¶ Slices the
self
tensor along the selected dimension at the given index. This function returns a tensor with the given dimension removed.Note
select()
is equivalent to slicing. For example,tensor.select(0, index)
is equivalent totensor[index]
andtensor.select(2, index)
is equivalent totensor[:,:,index]
.

set_
(source=None, storage_offset=0, size=None, stride=None) → Tensor¶ Sets the underlying storage, size, and strides. If
source
is a tensor,self
tensor will share the same storage and have the same size and strides assource
. Changes to elements in one tensor will be reflected in the other.If
source
is aStorage
, the method sets the underlying storage, offset, size, and stride.
Moves the underlying storage to shared memory.
This is a noop if the underlying storage is already in shared memory and for CUDA tensors. Tensors in shared memory cannot be resized.

sigmoid
() → Tensor¶ See
torch.sigmoid()

sign
() → Tensor¶ See
torch.sign()

sin
() → Tensor¶ See
torch.sin()

sinh
() → Tensor¶ See
torch.sinh()

size
() → torch.Size¶ Returns the size of the
self
tensor. The returned value is a subclass oftuple
.Example:
>>> torch.empty(3, 4, 5).size() torch.Size([3, 4, 5])

slogdet
() > (Tensor, Tensor)¶ See
torch.slogdet()

solve
(A) → Tensor, Tensor¶ See
torch.solve()

sort
(dim=1, descending=False) > (Tensor, LongTensor)¶ See
torch.sort()

split
(split_size, dim=0)[source]¶ See
torch.split()

sparse_mask
(input, mask) → Tensor¶ Returns a new SparseTensor with values from Tensor
input
filtered by indices ofmask
and values are ignored.input
andmask
must have the same shape. Parameters
input (Tensor) – an input Tensor
mask (SparseTensor) – a SparseTensor which we filter
input
based on its indices
Example:
>>> nnz = 5 >>> dims = [5, 5, 2, 2] >>> I = torch.cat([torch.randint(0, dims[0], size=(nnz,)), torch.randint(0, dims[1], size=(nnz,))], 0).reshape(2, nnz) >>> V = torch.randn(nnz, dims[2], dims[3]) >>> size = torch.Size(dims) >>> S = torch.sparse_coo_tensor(I, V, size).coalesce() >>> D = torch.randn(dims) >>> D.sparse_mask(S) tensor(indices=tensor([[0, 0, 0, 2], [0, 1, 4, 3]]), values=tensor([[[ 1.6550, 0.2397], [0.1611, 0.0779]], [[ 0.2326, 1.0558], [ 1.4711, 1.9678]], [[0.5138, 0.0411], [ 1.9417, 0.5158]], [[ 0.0793, 0.0036], [0.2569, 0.1055]]]), size=(5, 5, 2, 2), nnz=4, layout=torch.sparse_coo)

sparse_dim
() → int¶ If
self
is a sparse COO tensor (i.e., withtorch.sparse_coo
layout), this returns a the number of sparse dimensions. Otherwise, this throws an error.See also
Tensor.dense_dim()
.

sqrt
() → Tensor¶ See
torch.sqrt()

squeeze
(dim=None) → Tensor¶ See
torch.squeeze()

std
(dim=None, unbiased=True, keepdim=False) → Tensor¶ See
torch.std()

stft
(n_fft, hop_length=None, win_length=None, window=None, center=True, pad_mode='reflect', normalized=False, onesided=True)[source]¶ See
torch.stft()
Warning
This function changed signature at version 0.4.1. Calling with the previous signature may cause error or return incorrect result.

storage
() → torch.Storage¶ Returns the underlying storage.

storage_offset
() → int¶ Returns
self
tensor’s offset in the underlying storage in terms of number of storage elements (not bytes).Example:
>>> x = torch.tensor([1, 2, 3, 4, 5]) >>> x.storage_offset() 0 >>> x[3:].storage_offset() 3

storage_type
() → type¶ Returns the type of the underlying storage.

stride
(dim) → tuple or int¶ Returns the stride of
self
tensor.Stride is the jump necessary to go from one element to the next one in the specified dimension
dim
. A tuple of all strides is returned when no argument is passed in. Otherwise, an integer value is returned as the stride in the particular dimensiondim
. Parameters
dim (int, optional) – the desired dimension in which stride is required
Example:
>>> x = torch.tensor([[1, 2, 3, 4, 5], [6, 7, 8, 9, 10]]) >>> x.stride() (5, 1) >>>x.stride(0) 5 >>> x.stride(1) 1

sub
(value, other) → Tensor¶ Subtracts a scalar or tensor from
self
tensor. If bothvalue
andother
are specified, each element ofother
is scaled byvalue
before being used.When
other
is a tensor, the shape ofother
must be broadcastable with the shape of the underlying tensor.

sum
(dim=None, keepdim=False, dtype=None) → Tensor¶ See
torch.sum()

sum_to_size
(*size) → Tensor¶ Sum
this
tensor tosize
.size
must be broadcastable tothis
tensor size. :param other: The result tensor has the same sizeas
other
.

svd
(some=True, compute_uv=True) > (Tensor, Tensor, Tensor)¶ See
torch.svd()

symeig
(eigenvectors=False, upper=True) > (Tensor, Tensor)¶ See
torch.symeig()

to
(*args, **kwargs) → Tensor¶ Performs Tensor dtype and/or device conversion. A
torch.dtype
andtorch.device
are inferred from the arguments ofself.to(*args, **kwargs)
.Note
If the
self
Tensor already has the correcttorch.dtype
andtorch.device
, thenself
is returned. Otherwise, the returned tensor is a copy ofself
with the desiredtorch.dtype
andtorch.device
.Here are the ways to call
to
:
to
(dtype, non_blocking=False, copy=False) → Tensor Returns a Tensor with the specified
dtype

to
(device=None, dtype=None, non_blocking=False, copy=False) → Tensor Returns a Tensor with the specified
device
and (optional)dtype
. Ifdtype
isNone
it is inferred to beself.dtype
. Whennon_blocking
, tries to convert asynchronously with respect to the host if possible, e.g., converting a CPU Tensor with pinned memory to a CUDA Tensor. Whencopy
is set, a new Tensor is created even when the Tensor already matches the desired conversion.

to
(other, non_blocking=False, copy=False) → Tensor Returns a Tensor with same
torch.dtype
andtorch.device
as the Tensorother
. Whennon_blocking
, tries to convert asynchronously with respect to the host if possible, e.g., converting a CPU Tensor with pinned memory to a CUDA Tensor. Whencopy
is set, a new Tensor is created even when the Tensor already matches the desired conversion.
Example:
>>> tensor = torch.randn(2, 2) # Initially dtype=float32, device=cpu >>> tensor.to(torch.float64) tensor([[0.5044, 0.0005], [ 0.3310, 0.0584]], dtype=torch.float64) >>> cuda0 = torch.device('cuda:0') >>> tensor.to(cuda0) tensor([[0.5044, 0.0005], [ 0.3310, 0.0584]], device='cuda:0') >>> tensor.to(cuda0, dtype=torch.float64) tensor([[0.5044, 0.0005], [ 0.3310, 0.0584]], dtype=torch.float64, device='cuda:0') >>> other = torch.randn((), dtype=torch.float64, device=cuda0) >>> tensor.to(other, non_blocking=True) tensor([[0.5044, 0.0005], [ 0.3310, 0.0584]], dtype=torch.float64, device='cuda:0')


to_mkldnn
() → Tensor¶ Returns a copy of the tensor in
torch.mkldnn
layout.

take
(indices) → Tensor¶ See
torch.take()

tan
() → Tensor¶ See
torch.tan()

tanh
() → Tensor¶ See
torch.tanh()

tolist
()¶ ” tolist() > list or number
Returns the tensor as a (nested) list. For scalars, a standard Python number is returned, just like with
item()
. Tensors are automatically moved to the CPU first if necessary.This operation is not differentiable.
Examples:
>>> a = torch.randn(2, 2) >>> a.tolist() [[0.012766935862600803, 0.5415473580360413], [0.08909505605697632, 0.7729271650314331]] >>> a[0,0].tolist() 0.012766935862600803

topk
(k, dim=None, largest=True, sorted=True) > (Tensor, LongTensor)¶ See
torch.topk()

to_sparse
(sparseDims) → Tensor¶ Returns a sparse copy of the tensor. PyTorch supports sparse tensors in coordinate format.
 Parameters
sparseDims (int, optional) – the number of sparse dimensions to include in the new sparse tensor
Example:
>>> d = torch.tensor([[0, 0, 0], [9, 0, 10], [0, 0, 0]]) >>> d tensor([[ 0, 0, 0], [ 9, 0, 10], [ 0, 0, 0]]) >>> d.to_sparse() tensor(indices=tensor([[1, 1], [0, 2]]), values=tensor([ 9, 10]), size=(3, 3), nnz=2, layout=torch.sparse_coo) >>> d.to_sparse(1) tensor(indices=tensor([[1]]), values=tensor([[ 9, 0, 10]]), size=(3, 3), nnz=1, layout=torch.sparse_coo)

trace
() → Tensor¶ See
torch.trace()

transpose
(dim0, dim1) → Tensor¶

transpose_
(dim0, dim1) → Tensor¶ Inplace version of
transpose()

triangular_solve
(A, upper=True, transpose=False, unitriangular=False) > (Tensor, Tensor)¶

tril
(k=0) → Tensor¶ See
torch.tril()

triu
(k=0) → Tensor¶ See
torch.triu()

trunc
() → Tensor¶ See
torch.trunc()

type
(dtype=None, non_blocking=False, **kwargs) → str or Tensor¶ Returns the type if dtype is not provided, else casts this object to the specified type.
If this is already of the correct type, no copy is performed and the original object is returned.
 Parameters
dtype (type or string) – The desired type
non_blocking (bool) – If
True
, and the source is in pinned memory and destination is on the GPU or vice versa, the copy is performed asynchronously with respect to the host. Otherwise, the argument has no effect.**kwargs – For compatibility, may contain the key
async
in place of thenon_blocking
argument. Theasync
arg is deprecated.

type_as
(tensor) → Tensor¶ Returns this tensor cast to the type of the given tensor.
This is a noop if the tensor is already of the correct type. This is equivalent to
self.type(tensor.type())
 Parameters
tensor (Tensor) – the tensor which has the desired type

unbind
(dim=0) → seq¶ See
torch.unbind()

unfold
(dimension, size, step) → Tensor¶ Returns a tensor which contains all slices of size
size
fromself
tensor in the dimensiondimension
.Step between two slices is given by
step
.If sizedim is the size of dimension
dimension
forself
, the size of dimensiondimension
in the returned tensor will be (sizedim  size) / step + 1.An additional dimension of size
size
is appended in the returned tensor. Parameters
Example:
>>> x = torch.arange(1., 8) >>> x tensor([ 1., 2., 3., 4., 5., 6., 7.]) >>> x.unfold(0, 2, 1) tensor([[ 1., 2.], [ 2., 3.], [ 3., 4.], [ 4., 5.], [ 5., 6.], [ 6., 7.]]) >>> x.unfold(0, 2, 2) tensor([[ 1., 2.], [ 3., 4.], [ 5., 6.]])

uniform_
(from=0, to=1) → Tensor¶ Fills
self
tensor with numbers sampled from the continuous uniform distribution:\[P(x) = \dfrac{1}{\text{to}  \text{from}} \]

unique
(sorted=True, return_inverse=False, return_counts=False, dim=None)[source]¶ Returns the unique elements of the input tensor.
See
torch.unique()

unique_consecutive
(return_inverse=False, return_counts=False, dim=None)[source]¶ Eliminates all but the first element from every consecutive group of equivalent elements.

unsqueeze
(dim) → Tensor¶

unsqueeze_
(dim) → Tensor¶ Inplace version of
unsqueeze()

values
() → Tensor¶ If
self
is a sparse COO tensor (i.e., withtorch.sparse_coo
layout), this returns a view of the contained values tensor. Otherwise, this throws an error.See also
Tensor.indices()
.Note
This method can only be called on a coalesced sparse tensor. See
Tensor.coalesce()
for details.

var
(dim=None, unbiased=True, keepdim=False) → Tensor¶ See
torch.var()

view
(*shape) → Tensor¶ Returns a new tensor with the same data as the
self
tensor but of a differentshape
.The returned tensor shares the same data and must have the same number of elements, but may have a different size. For a tensor to be viewed, the new view size must be compatible with its original size and stride, i.e., each new view dimension must either be a subspace of an original dimension, or only span across original dimensions \(d, d+1, \dots, d+k\) that satisfy the following contiguitylike condition that \(\forall i = 0, \dots, k1\),
\[\text{stride}[i] = \text{stride}[i+1] \times \text{size}[i+1]\]Otherwise,
contiguous()
needs to be called before the tensor can be viewed. See also:reshape()
, which returns a view if the shapes are compatible, and copies (equivalent to callingcontiguous()
) otherwise. Parameters
shape (torch.Size or int...) – the desired size
Example:
>>> x = torch.randn(4, 4) >>> x.size() torch.Size([4, 4]) >>> y = x.view(16) >>> y.size() torch.Size([16]) >>> z = x.view(1, 8) # the size 1 is inferred from other dimensions >>> z.size() torch.Size([2, 8]) >>> a = torch.randn(1, 2, 3, 4) >>> a.size() torch.Size([1, 2, 3, 4]) >>> b = a.transpose(1, 2) # Swaps 2nd and 3rd dimension >>> b.size() torch.Size([1, 3, 2, 4]) >>> c = a.view(1, 3, 2, 4) # Does not change tensor layout in memory >>> c.size() torch.Size([1, 3, 2, 4]) >>> torch.equal(b, c) False

view_as
(other) → Tensor¶ View this tensor as the same size as
other
.self.view_as(other)
is equivalent toself.view(other.size())
.Please see
view()
for more information aboutview
. Parameters
other (
torch.Tensor
) – The result tensor has the same size asother
.

where
(condition, y) → Tensor¶ self.where(condition, y)
is equivalent totorch.where(condition, self, y)
. Seetorch.where()

zero_
() → Tensor¶ Fills
self
tensor with zeros.

class
torch.
ByteTensor
¶ The following methods are unique to
torch.ByteTensor
.
all
()¶ 
all
() → bool
Returns True if all elements in the tensor are nonzero, False otherwise.
Example:
>>> a = torch.randn(1, 3).byte() % 2 >>> a tensor([[1, 0, 0]], dtype=torch.uint8) >>> a.all() tensor(0, dtype=torch.uint8)

all
(dim, keepdim=False, out=None) → Tensor
Returns True if all elements in each row of the tensor in the given dimension
dim
are nonzero, False otherwise.If
keepdim
isTrue
, the output tensor is of the same size asinput
except in the dimensiondim
where it is of size 1. Otherwise,dim
is squeezed (seetorch.squeeze()
), resulting in the output tensor having 1 fewer dimension thaninput
. Parameters
Example:
>>> a = torch.randn(4, 2).byte() % 2 >>> a tensor([[0, 0], [0, 0], [0, 1], [1, 1]], dtype=torch.uint8) >>> a.all(dim=1) tensor([0, 0, 0, 1], dtype=torch.uint8)


any
()¶ 
any
() → bool
Returns True if any elements in the tensor are nonzero, False otherwise.
Example:
>>> a = torch.randn(1, 3).byte() % 2 >>> a tensor([[0, 0, 1]], dtype=torch.uint8) >>> a.any() tensor(1, dtype=torch.uint8)

any
(dim, keepdim=False, out=None) → Tensor
Returns True if any elements in each row of the tensor in the given dimension
dim
are nonzero, False otherwise.If
keepdim
isTrue
, the output tensor is of the same size asinput
except in the dimensiondim
where it is of size 1. Otherwise,dim
is squeezed (seetorch.squeeze()
), resulting in the output tensor having 1 fewer dimension thaninput
. Parameters
Example:
>>> a = torch.randn(4, 2).byte() % 2 >>> a tensor([[1, 0], [0, 0], [0, 1], [0, 0]], dtype=torch.uint8) >>> a.any(dim=1) tensor([1, 0, 1, 0], dtype=torch.uint8)

