Expand description
This crate is dynamic graph based tensor library
Re-exports§
Modules§
- arch_
simd - A module defines a set of vector types using stdsimd
- ops
- a module contains all the Tensor operations. include the CPU and GPU operations
- random
- A module contains random number generation operations
- shape_
manipulate - A module contains shape manipulation operations
- tensor
- a module that wrap the _Tensor struct
- tensor_
base - a module that defines the _Tensor struct
- to_
tensor - a module that contains the implementation of the
Intotrait for the_Tensorstruct. - traits
- A module defines a set of traits for vector
- utils
- A module defines a set of utils for vector
Macros§
- select
- parse the input and generate the corresponding slice
Structs§
- Backend
- backend of tensor
- Cpu
- Cpu backend
- Data
Loader - Shape
- Represents the shape of a multi-dimensional structure, such as a tensor or an array.
- Strides
- Represents the strides of a multi-dimensional structure, such as a tensor or an array.
- Tensor
Loader - Tensor
Saver
Enums§
- Compression
Algo - Endian
- Tensor
Error - Base error type for all tensor operations
Traits§
- Advanced
Ops - A trait contains advance operations
- Allocator
- traits for the allocator
- Allocator
Output Retrive - traits for the allocator output retrive
- Backend
Ty - backend id trait
- BitWise
Out - this trait is used to perform bitwise operations
- Buffer
- trait for buffer
- Cast
- A trait for converting a scalar into another scalar type.
- Common
Bounds - Common bounds for primitive types
- Concat
- trait for concat
- Conv
- trait for conv operations
- Conv
Diff - trait for differentiable conv operations
- Cum
- A trait for accumulative operations.
- Cumulative
Ops - A trait for cumulative operations
- Eval
- this trait is used to perform evaluation operations
- Eval
Reduce - A trait for tensor reduction operations, the output must be a boolean tensor.
- FFTOps
- A trait for Fast Fourier Transform (FFT) operations.
- Float
BinOps - A trait for binary operations on tensors.
- Float
OutBinary - this trait is used to perform type promotion in dynamic graph
- Float
OutBinary Promote - this trait is used to perform type promotion for float out binary operations
- Float
OutPooling - trait for pooling that the output type is the same as the input type
- Float
OutUnary - This trait is used to perform various unary floating-point operations.
- Float
OutUnary Promote - this trait is used to promote the float out unary trait to the output type
- Float
Reduce - A trait for tensor reduction operations, the output must be a floating-point tensor.
- Float
Unary Ops - A trait for tensor unary operations, the output must be a floating point tensor
- From
Safe Tensors - HardMax
- A trait for hardmax
- Index
Reduce - A trait typically for argmax and argmin functions.
- IntoVec
- a trait to convert a vector to another vector
- Iter
GetSet - A trait for getting and setting values from an iterator.
- Iter
GetSet Simd - A trait for getting and setting values from an simd iterator
- Load
- Matmul
- A trait for matrix multiplication operations on tensors.
- Meta
Load - Normal
BinOps - A trait for binary operations on tensors.
- Normal
Eval Reduce - A trait for tensor reduction operations, the output must remain the same tensor type.
- Normal
Out - this trait is used to perform normal operations that don’t require type promotion
- Normal
OutPromote - this trait is used to perform type promotion for normal out operations
- Normal
OutUnary - this trait is used to perform normal unary operations that don’t require type promotion
- Normal
Pooling - trait for pooling that the output type is the same as the input type
- Normal
Reduce - A trait for normal tensor reduction operations.
- Normal
Uary Ops - A trait for unary operations, the output must be the same type as the input.
- ParStrided
Iterator Simd - A trait for performing single thread simd iteration over an iterator.
- ParStrided
Iterator Simd Zip - A trait to zip two parallel iterators together.
- ParStrided
Iterator Zip - A trait to zip two parallel iterators together.
- Random
- A trait for generating random numbers.
- Random
Int - A trait for generating random integers.
- Save
- Shape
Manipulate - A trait for manipulating the shape of a tensor.
- Shape
Manipulator - A trait for performing shape manipulation on an iterator.
- Shrinkage
- A trait for shrinkage
- Slice
- trait for slicing tensor
- Strided
Iterator - A trait for performing single thread iteration over an iterator.
- Strided
Iterator Map - A trait to map a function on the elements of an iterator.
- Strided
Iterator Simd - A trait for performing single thread simd iteration over an iterator.
- Strided
Iterator Zip - A trait to zip two iterators together.
- Strided
Simd Iterator Zip - A trait to zip two simd iterators together.
- Tensor
Alloc - A trait for tensor memory allocation, this trait only used when we work with generic type
- Tensor
Cmp - A trait for tensor comparison operations
- Tensor
Creator - A trait defines a set of functions to create tensors.
- Tensor
Dot - A trait for tensor dot operations on tensors.
- Tensor
Info - A trait for getting information of a Tensor
- Tensor
Iterator - A trait for converting a tensor into an iterator.
- Tensor
Like - A trait for let the object like a tensor
- Tensor
Where - A trait for tensor where
- Type
Common - common trait for all data types
- VecTrait
- common trait for all vector types
- Window
Ops - A trait contains window operations
Functions§
- binary_
with_ out - Perform binary operation with output tensor
- get_
num_ threads - Get the global number of threads
- parse_
header_ compressed - resize_
cpu_ lru_ cache - resize the lru cache of the cpu allocator
- set_
display_ elements - Set the left and right elements to display for each dimension
- set_
display_ precision - Set the Tensor display precision
- set_
num_ threads - Set the global number of threads
- set_
seed - Set the seed for random number generation