How to Use the candle ML Framework in Rust

Ai Sovereignty
2026-04-17T12:05:42+00:00
Use the candle crate to build and run deep learning models in Rust by adding dependencies and initializing tensors.

There is no official "candle ML framework" in Rust; you are likely looking for the candle crate, a deep learning library. Add it to your project and initialize a model to start using it.

[dependencies]
candle-core = "0.8"
candle-nn = "0.8"

use candle_core as candle;
use candle_nn::Module;

fn main() -> candle::Result<()> {
    let device = candle::Device::Cpu;
    let model = candle_nn::linear(10, 5, &device)?;
    let input = candle::Tensor::randn(0., 1., (1, 10), &device)?;
    let output = model.forward(&input)?;
    Ok(())
}