Install RUMUS

Get up and running with RUMUS in minutes. Select your configuration below to see the right installation commands.

Prerequisites

Rust

Install via rustup.rs

1.75+
bash
"token">-function">curl "token-keyword">--proto '=https' "token-keyword">--tlsv1.2 "token-keyword">-sSf https://sh.rustup.rs | sh

GPU support (optional)

WGPU requires a compatible graphics driver

Vulkan/Metal/DX12
bash
"token">-comment"># macOS: Metal is built"token-keyword">-in
"token">-comment"># Linux: Install Vulkan SDK
"token">-comment"># Windows: DirectX 12 or Vulkan

Configuration

bash
"token">-comment"># Add RUMUS to your project
"token">-function">cargo add rumus

"token">-comment"># Or add manually to Cargo.toml
"token">-comment"># [dependencies]
"token">-comment"># rumus = "0.1"
bash
"token">-comment"># Build your project (CPU"token-keyword">-only)
"token">-function">cargo build

"token">-comment"># Run tests
"token">-function">cargo test

Verify Installation

Create a new file and run it to verify everything is working:

src/main.rs
rust
use rumus::type">Tensor;
use rumus::nn::{self, type">Linear, type">Module};

fn main() {
    "token-comment">// Create a tensor
    let x = type">Tensor::new(vec![1.0, 2.0, 3.0, 4.0], vec![2, 2]);
    println!("type">Tensor shape: {:?}", x.shape());

    "token-comment">// Create a linear layer
    let layer = type">Linear::new(2, 3);
    let out = layer.forward(&x);
    println!("Output shape: {:?}", out.shape());

    println!("RUMUS is working!");
}

Expected output

Tensor shape: [2, 2]
Output shape: [2, 3]
RUMUS is working!

Dependencies

RUMUS has minimal dependencies — the core framework has no heavy runtime requirements:

PackageVersionPurpose
parking_lot0.12RwLock with zero-copy data access guards
safetensors0.4Safe model serialization format
bytemuck1.14Type-safe f32/bytes casting (zero unsafe)
wgpu (gpu feature)24Cross-platform GPU compute via WebGPU
pollster (gpu feature)0.4Minimal async runtime for GPU init

Next Steps

Getting Started Guide

Build your first neural network with RUMUS

Tensor Operations

Learn about tensor creation, views, and arithmetic

Neural Network Modules

Explore layers, activations, and the Module system

GPU Acceleration

Move your training to the GPU with WGPU