xn 0.1.1

Another minimalist deep-learning framework optimized for inference.
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
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70

#[cfg(feature = "accelerate")]
mod accelerate;

pub mod backend;
pub mod cpu_backend;
pub mod display;
pub mod dtype;
pub mod error;
pub mod inplace_ops;
pub mod models;
pub mod nn;
pub mod ops;
pub mod safetensors;
pub mod shape;
pub mod streaming;
pub mod tensor;
pub mod tensor_view;
pub mod utils;

pub use backend::Backend;
pub use dtype::{DType, WithDType, WithDTypeF};
pub use error::{Error, Result};
pub use shape::{D, Dim, Shape};
pub use tensor::{Tensor, TypedTensor};
pub use tensor_view::{TensorOrView, TensorView};
pub use utils::{get_num_cpus, get_num_threads, set_num_threads};

#[derive(Clone, Copy, Debug, PartialEq, Eq, Hash)]
pub struct CpuDevice;
pub type CpuTensor<T> = Tensor<T, CpuDevice>;

pub const CPU: CpuDevice = CpuDevice;

pub(crate) use inplace_ops::{BinaryOp, UnaryOp};

#[cfg(feature = "cuda")]
pub mod cuda_backend;
#[cfg(feature = "cuda")]
pub mod cuda_kernels;
#[cfg(feature = "cuda")]
pub use cuda_backend::Device as CudaDevice;

pub fn with_avx() -> bool {
    cfg!(target_feature = "avx")
}

pub fn with_neon() -> bool {
    cfg!(target_feature = "neon")
}

pub fn with_simd128() -> bool {
    cfg!(target_feature = "simd128")
}

pub fn with_f16c() -> bool {
    cfg!(target_feature = "f16c")
}

pub trait Module {
    fn forward<T: WithDType, B: Backend>(&self, xs: &Tensor<T, B>) -> Result<Tensor<T, B>>;
}

impl<M: Module> Module for Option<&M> {
    fn forward<T: WithDType, B: Backend>(&self, xs: &Tensor<T, B>) -> Result<Tensor<T, B>> {
        match self {
            None => Ok(xs.clone()),
            Some(m) => m.forward(xs),
        }
    }
}