[][src]Crate autograph

Modules

datasets
nn
utils

Structs

Conv2dArgs

Builder struct for 2D Convolution functions
Additional features may be added

Cpu

Threadsafe wrapper for oneDNN engine and stream

Pool2dArgs

Similar to Conv2dArgs, but for pooling functions

TensorBase

The core data type of Autograph, which abstracts over data location and representation. Data must be loaded into tensors before operations can be run.

Enums

Device

Device is an enum that is used to select whether to store and execute operations on the cpu or a gpu.

Traits

Data

Main trait for Tensor S generic parameter, similar to ndarray::Data. Elem indicates that the Tensor stores that datatype.

DataMut

Trait for Tensors which can borrow their data mutably

DataOwned

Trait for Tensors that can be constructed, and do not have references (ie not a View)

DataRef

Trait for Tensors which can borrow their data immutably

Into2d

Convenience trait to allow passing either usize or [usize, usize] to functions
ie 1.into_2d() == [1, 1]

Num

Num is a trait for all data types that Tensor can store, it cannot be implemented for additional types

Unsigned

Unsigned is a trait for types which can be treated as an index, ie converted to usize

Type Definitions

ArcTensor

Tensor which has threadsafe shared immutable access without a lifetime

ArcTensor2
ArcTensor4
ArcTensorD
RwReadTensor

Represents an immutable borrow of a RwTensor, acts like a TensorView

RwTensor

Tensor which allows for either shared immutable access or exclusive mutable access

RwTensor0
RwTensor1
RwTensor2
RwTensor3
RwTensor4
RwTensorD
RwWriteTensor

Represents a mutable borrow of a RwTensor, acts like a TensorViewMut

Tensor

Tensor which has exclusive ownership of its data

Tensor0
Tensor1
Tensor2
Tensor4
TensorD
TensorView

Tensor which has an immutable (shared) borrow of its data

TensorView1
TensorView2
TensorView4
TensorViewD
TensorViewMut

Tensor which has a mutable (exclusive) borrow of its data

TensorViewMut0
TensorViewMut1
TensorViewMut2
TensorViewMut4

Attribute Macros

impl_forward

Derive Macros

Layer