#opencl #csv #numpy #data #cuda #binary-file

RayBNN_DataLoader

Read CSV, numpy, and binary files to Rust vectors of f16, f32, f64, u8, u16, u32, u64, i8, i16, i32, i64

7 releases (3 stable)

2.0.3 Jun 18, 2024
2.0.2 Jun 5, 2024
0.1.3 Oct 5, 2023
0.1.2 Sep 5, 2023

#87 in Science

Download history 29/week @ 2024-06-21 4/week @ 2024-06-28 12/week @ 2024-07-05 8/week @ 2024-07-12 7/week @ 2024-07-26 5/week @ 2024-08-02 3/week @ 2024-08-09 2/week @ 2024-08-16 6/week @ 2024-08-23 5/week @ 2024-08-30 4/week @ 2024-09-13 27/week @ 2024-09-20 19/week @ 2024-09-27 2/week @ 2024-10-04

480 downloads per month
Used in 6 crates

GPL-3.0-only

105KB
322 lines

RayBNN_DataLoader

Data Loader for RayBNN

Read CSV, numpy, and binary files to Rust vectors of f16, f32, f64, u8, u16, u32, u64, i8, i16, i32, i64

Read CSV, numpy, and binary files to Arrayfire GPU arrays of f16, f32, f64, u8, u16, u32, u64, i8, i16, i32, i64

Install Arrayfire

Install the Arrayfire 3.9.0 binaries at https://arrayfire.com/binaries/

or build from source https://github.com/arrayfire/arrayfire/wiki/Getting-ArrayFire

Add to Cargo.toml

arrayfire = { version = "3.8.1", package = "arrayfire_fork" }
rayon = "1.10.0"
num = "0.4.3"
num-traits = "0.2.19"
half = { version = "2.4.1" , features = ["num-traits"] }
npyz = "0.8.3"
RayBNN_DataLoader = "2.0.3"

List of Examples

Read a CSV file to a floating point 64 bit CPU Vector

The vector is completely flat

let (mut cpu_vector,metadata) = RayBNN_DataLoader::Dataset::CSV::file_to_vec_cpu::<f64>(
    "./test_data/read_test.dat"
);

Read a CSV file to a integer 64 bit CPU Vector

The vector is completely flat

let (mut cpu_vector,metadata) = RayBNN_DataLoader::Dataset::CSV::file_to_vec_cpu::<i64>(
    "./test_data/read_test2.dat"
);

Read a CSV file to a floating point 64 bit arrayfire

The array is 2D existing in GPU or OpenCL

let read_test = RayBNN_DataLoader::Dataset::CSV::file_to_arrayfire::<f64>(
    "./test_data/read_test.dat"
);

Read a CSV file to a floating point 64 bit HashMap

let hashdata = RayBNN_DataLoader::Dataset::CSV::file_to_hash_cpu::<f64>(
    "./test_data/dataloader.csv",
    7,
    5
);

Read a CSV file to a floating point 64 bit HashMap with Arrayfire

let arr_dims = arrayfire::Dim4::new(&[5, 7, 1, 1]);
let (hashdata,metadata) = RayBNN_DataLoader::Dataset::CSV::file_to_hash_arrayfire::<f64>(
    "./test_data/dataloader.csv",
    arr_dims,
);

Write a float 32 bit CPU vector to CSV file

let mut metadata: HashMap<String,u64> = HashMap::new();

metadata.insert("dim0", 11);
metadata.insert("dim1", 3);

RayBNN_DataLoader::Dataset::CSV::write_vec_cpu_to_csv::<f32>(
    "./randvec2.csv",
    &randvec,
    &metadata
);

Write a float 64 bit arrayfire to CSV file

RayBNN_DataLoader::Dataset::CSV::write_arrayfire_to_csv::<f64>(
    "./randvec.csv",
    &arr
);

Read YAML Model Information

let mut modeldata_string:  HashMap<String, String> = HashMap::new();
let mut modeldata_float:  HashMap<String, f64> = HashMap::new();
let mut modeldata_int:  HashMap<String, u64> = HashMap::new();

RayBNN_DataLoader::Model::YAML::read(
    "./test_data/test.yaml",

    &mut modeldata_string,
    &mut modeldata_float,
    &mut modeldata_int,
);

Write YAML Model Information

RayBNN_DataLoader::Model::YAML::write(
    "./sample.yaml",

    &modeldata_string,
    &modeldata_float,
    &modeldata_int,
);

Dependencies

~7MB
~139K SLoC