dfdx 0.9.0

Ergonomic auto differentiation in Rust, with pytorch like apis.
Documentation 
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
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46

//! We use [std::rc::Rc] instead of [Box] here to reduce allocations when tensors are duplicated/cloned.
//!
//! See [#62](https://github.com/coreylowman/dfdx/issues/62) for more discussion.

use crate::{gradients::NoneTape, unique_id::UniqueId};

/// A 0d [super::Tensor] with shape (). Backed by data `f32`.
#[derive(Debug)]
pub struct Tensor0D<Tape = NoneTape> {
    pub(crate) id: UniqueId,
    pub(crate) data: std::rc::Rc<f32>,
    pub(crate) tape: Tape,
}

/// A 1d [super::Tensor] with shape (M, ). Backed by data `[f32; M]`.
#[derive(Debug)]
pub struct Tensor1D<const N: usize, Tape = NoneTape> {
    pub(crate) id: UniqueId,
    pub(crate) data: std::rc::Rc<[f32; N]>,
    pub(crate) tape: Tape,
}

/// A 2d [super::Tensor] with shape (M, N). Backed by data `[[f32; N]; M]`.
#[derive(Debug)]
pub struct Tensor2D<const M: usize, const N: usize, Tape = NoneTape> {
    pub(crate) id: UniqueId,
    pub(crate) data: std::rc::Rc<[[f32; N]; M]>,
    pub(crate) tape: Tape,
}

/// A 3d [super::Tensor] with shape (M, N, O). Backed by data `[[[f32; O]; N]; M]`.
#[derive(Debug)]
pub struct Tensor3D<const M: usize, const N: usize, const O: usize, Tape = NoneTape> {
    pub(crate) id: UniqueId,
    pub(crate) data: std::rc::Rc<[[[f32; O]; N]; M]>,
    pub(crate) tape: Tape,
}

/// A 4d [super::Tensor] with shape (M, N, O, P). Backed by data `[[[[f32; P]; O]; N]; M]`.
#[derive(Debug)]
pub struct Tensor4D<const M: usize, const N: usize, const O: usize, const P: usize, Tape = NoneTape>
{
    pub(crate) id: UniqueId,
    pub(crate) data: std::rc::Rc<[[[[f32; P]; O]; N]; M]>,
    pub(crate) tape: Tape,
}