137 releases

0.21.9 Jan 8, 2025
0.21.8 Dec 5, 2024
0.21.7 Sep 23, 2024
0.21.6 Jul 24, 2024
0.1.1 Nov 2, 2018

#22 in #self-contained

Download history 8257/week @ 2024-10-09 9077/week @ 2024-10-16 10505/week @ 2024-10-23 4666/week @ 2024-10-30 3248/week @ 2024-11-06 2976/week @ 2024-11-13 4887/week @ 2024-11-20 3755/week @ 2024-11-27 3991/week @ 2024-12-04 4060/week @ 2024-12-11 2795/week @ 2024-12-18 1527/week @ 2024-12-25 3095/week @ 2025-01-01 4409/week @ 2025-01-08 4713/week @ 2025-01-15 3546/week @ 2025-01-22

16,074 downloads per month
Used in 37 crates (9 directly)

MIT/Apache

2MB
51K SLoC

Rust 42K SLoC // 0.0% comments Templ 9K SLoC // 0.1% comments GNU Style Assembly 12 SLoC // 0.2% comments

Tract

Tiny, no-nonsense, self contained, portable TensorFlow and ONNX inference.

Example

use tract_core::internal::*;

// build a simple model that just add 3 to each input component
let mut model = TypedModel::default();

let input_fact = f32::fact(&[3]);
let input = model.add_source("input", input_fact).unwrap();
let three = model.add_const("three".to_string(), tensor1(&[3f32])).unwrap();
let add = model.wire_node("add".to_string(),
    tract_core::ops::math::add(),
    [input, three].as_ref()
    ).unwrap();

model.auto_outputs().unwrap();

// We build an execution plan. Default inputs and outputs are inferred from
// the model graph.
let plan = SimplePlan::new(&model).unwrap();

// run the computation.
let input = tensor1(&[1.0f32, 2.5, 5.0]);
let mut outputs = plan.run(tvec![input.into()]).unwrap();

// take the first and only output tensor
let mut tensor = outputs.pop().unwrap();

assert_eq!(tensor, tensor1(&[4.0f32, 5.5, 8.0]).into());

While creating a model from Rust code is useful for testing the library, real-life use-cases will usually load a TensorFlow or ONNX model using tract-tensorflow or tract-onnx crates.

Dependencies

~13–25MB
~394K SLoC