Table of Contents

Installation Guide

{: .no_toc }

Complete guide to installing AiDotNet and its dependencies. {: .fs-6 .fw-300 }


Table of contents

{: .no_toc .text-delta }

  1. TOC {:toc}

Basic Installation

Using .NET CLI

dotnet add package AiDotNet

Using Package Manager Console

Install-Package AiDotNet

Using PackageReference

Add to your .csproj file:

<ItemGroup>
  <PackageReference Include="AiDotNet" Version="*" />
</ItemGroup>

Platform Support

Platform Status Notes
Windows x64 ✅ Full Recommended
Linux x64 ✅ Full Ubuntu 20.04+
macOS x64 ✅ Full macOS 11+
macOS ARM64 ✅ Full Apple Silicon

.NET Version Support

Version Status
.NET 8.0 ✅ Primary target
.NET 7.0 ✅ Supported
.NET 6.0 ✅ Supported
.NET Framework 4.6.2+ ✅ Supported

GPU Acceleration

NVIDIA CUDA

For GPU-accelerated training on NVIDIA GPUs:

  1. Install CUDA Toolkit 12.x+ (or 11.8+ minimum)

  2. Install cuDNN 9.x+ (or 8.6+ minimum)

    • Download from NVIDIA cuDNN
    • cuDNN 9.x recommended for compatibility with CUDA 12.x
  3. Verify installation

    nvcc --version
    

OpenCL

For AMD/Intel GPU support:

  1. Install OpenCL runtime for your GPU vendor
  2. Install CLBlast for optimized BLAS operations

Optional Packages

Model Serving

For production model serving:

dotnet add package AiDotNet.Serving

Dashboard

For training visualization:

dotnet add package AiDotNet.Dashboard

Verifying Installation

Create a simple test:

using AiDotNet;

Console.WriteLine("AiDotNet installed successfully!");

// Test basic functionality
var builder = new AiModelBuilder<double, double[], double>();
Console.WriteLine("AiModelBuilder created.");

Run it:

dotnet run

Troubleshooting

Package Not Found

Ensure you have the latest NuGet sources:

dotnet nuget list source

GPU Not Detected

  1. Verify CUDA installation: nvcc --version
  2. Check GPU visibility: nvidia-smi
  3. Ensure CUDA path is in environment variables

Build Errors

  1. Clear NuGet cache: dotnet nuget locals all --clear
  2. Restore packages: dotnet restore
  3. Rebuild: dotnet build

Next Steps