Preparing search index...
The search index is not available
webgpu-torch
webgpu-torch
GradientContext
Class GradientContext
Hierarchy
GradientContext
Indexable
[
key:
string
]:
any
Index
Constructors
constructor
Properties
inputs
With
Gradient
needs
Input
Gradient
saved
Tensors
Methods
save
For
Backward
Constructors
constructor
new
Gradient
Context
(
inputs
)
:
GradientContext
Parameters
inputs
:
FunctionInput
[]
Returns
GradientContext
Properties
inputs
With
Gradient
inputs
With
Gradient
:
(
null
|
Tensor
)
[]
needs
Input
Gradient
needs
Input
Gradient
:
boolean
[]
saved
Tensors
saved
Tensors
:
Tensor
[]
= []
Methods
save
For
Backward
save
For
Backward
(
...
tensors
)
:
void
Parameters
Rest
...
tensors
:
Tensor
[]
Returns
void
Settings
Member Visibility
Protected
Private
Inherited
External
Theme
OS
Light
Dark
On This Page
constructor
inputs
With
Gradient
needs
Input
Gradient
saved
Tensors
save
For
Backward
webgpu-
torch
init
Linear
Function
Nonlinear
Function
calculate
Fan
In
And
Fan
Out
kaiming
Uniform_
uniform_
nn
applications
DeepSDF
diffusers
Timestep
Embed
Sequential
UNet
Model
UNet
Model
Config
Avg
Pooling2d
Container
Conv2d
Conv
Transpose2d
Group
Norm
Linear
Module
Module
List
Parameter
ReLU
Sequential
SiLU
Sigmoid
Tanh
State
Dict
Device
Gradient
Context
Kernel
Tensor
Untyped
Storage
Kernel
Config
Spec
Kernel
Input
Spec
Kernel
Output
Spec
Kernel
Param
Spec
Kernel
Spec
Strided
Shape
ATyped
Array
Compiled
Expr
Device
Id
Device
Type
Deviceish
Dtype
Eval
Env
Expr
Code
Function
Input
Gradient
Function
Gradient
Function
Output
Kernel
Config
Kernel
Config
Input
Kernel
Key
Kernel
Params
Input
Memory
Format
Op
Spec
Op
Type
Shader
Type
Shape
Shapeish
Strides
Tensor
Array
Data
Tensor
Data
Tensor
Spec
abs
absolute
acos
acosh
add
all
any
arccos
arccosh
arcsin
arcsinh
arctan
arctan2
asin
asinh
atan
atan2
broadcast
Batched
Matmul
broadcast
Shapes
canonicalize
Dim
cat
ceil
check
clone
contiguous
Strided
Shape
conv2d
copysign
cos
cosh
count
Nonzero
default
Strides
deg2rad
div
divide
empty
exp
exp2
expm1
fix
flatten
floor
frac
gather
get
Shape
has
WebGPU
hypot
init
WebGPUAsync
ldexp
linear
log
log10
log1p
log2
logaddexp
logaddexp2
matmul
mean
mm
mul
multiply
neg
negative
norm
numel
ones
positive
pow
prod
rad2deg
reciprocal
relu
reshape
reshape
Batched
Matmul
round
rsqrt
save
Safetensors
Async
shape
Size
shapes
Are
Equal
sigmoid
sign
silu
sin
sinc
sinh
sqrt
square
squeeze
strided
Shape
Is
Contiguous
sub
subtract
sum
t
tan
tanh
tensor
trunc
unsqueeze
validate
Dim
Length
validate
Idx
view
xlogy
zeros
Generated using
TypeDoc