LLamaSharp 0.2.3

.NET 6.0 This package targets .NET 6.0. The package is compatible with this framework or higher. .NET Standard 2.0 This package targets .NET Standard 2.0. The package is compatible with this framework or higher.
There is a newer version of this package available.
See the version list below for details. 
dotnet add package LLamaSharp --version 0.2.3                

                    
                

            
NuGet\Install-Package LLamaSharp -Version 0.2.3
This command is intended to be used within the Package Manager Console in Visual Studio, as it uses the NuGet module's version of Install-Package. 
<PackageReference Include="LLamaSharp" Version="0.2.3" />
For projects that support PackageReference, copy this XML node into the project file to reference the package. 
paket add LLamaSharp --version 0.2.3                

                    
                

            
#r "nuget: LLamaSharp, 0.2.3"
#r directive can be used in F# Interactive and Polyglot Notebooks. Copy this into the interactive tool or source code of the script to reference the package. 
// Install LLamaSharp as a Cake Addin
#addin nuget:?package=LLamaSharp&version=0.2.3

// Install LLamaSharp as a Cake Tool
#tool nuget:?package=LLamaSharp&version=0.2.3

LLamaSharp - .NET Binding for llama.cpp

logo

The C#/.NET binding of llama.cpp. It provides APIs to inference the LLaMa Models and deploy it on native environment or Web. It works on both Windows and Linux and does NOT require compiling llama.cpp yourself.

  • Load and inference LLaMa models
  • Simple APIs for chat session
  • Quantize the model in C#/.NET
  • ASP.NET core integration
  • Native UI integration

Installation

Firstly, search LLamaSharp in nuget package manager and install it.

PM> Install-Package LLamaSharp

Then, search and install one of the following backends:

LLamaSharp.Backend.Cpu
LLamaSharp.Backend.Cuda11
LLamaSharp.Backend.Cuda12

The latest version of LLamaSharp and LLamaSharp.Backend may not always be the same. LLamaSharp.Backend follows up llama.cpp because sometimes the break change of it makes some model weights invalid. If you are not sure which version of backend to install, just install the latest version.

Note that version v0.2.1 has a package named LLamaSharp.Cpu. After v0.2.2 it will be dropped.

We publish the backend with cpu, cuda11 and cuda12 because they are the most popular ones. If none of them matches, please compile the llama.cpp from source and put the libllama under your project's output path. When building from source, please add -DBUILD_SHARED_LIBS=ON to enable the library generation.

Simple Benchmark

Currently it's only a simple benchmark to indicate that the performance of LLamaSharp is close to llama.cpp. Experiments run on a computer with Intel i7-12700, 3060Ti with 7B model. Note that the benchmark uses LLamaModel instead of LLamaModelV1.

Windows

  • llama.cpp: 2.98 words / second

  • LLamaSharp: 2.94 words / second

Usages

Model Inference and Chat Session

Currently, LLamaSharp provides two kinds of model, LLamaModelV1 and LLamaModel. Both of them works but LLamaModel is more recommended because it provides better alignment with the master branch of llama.cpp.

Besides, ChatSession makes it easier to wrap your own chat bot. The code below is a simple example. For all examples, please refer to Examples.


var model = new LLamaModel(new LLamaParams(model: "<Your path>", n_ctx: 512, repeat_penalty: 1.0f));
var session = new ChatSession<LLamaModel>(model).WithPromptFile("<Your prompt file path>")
                .WithAntiprompt(new string[] { "User:" });
Console.Write("\nUser:");
while (true)
{
    Console.ForegroundColor = ConsoleColor.Green;
    var question = Console.ReadLine();
    Console.ForegroundColor = ConsoleColor.White;
    var outputs = session.Chat(question); // It's simple to use the chat API.
    foreach (var output in outputs)
    {
        Console.Write(output);
    }
}
Quantization

The following example shows how to quantize the model. With LLamaSharp you needn't to compile c++ project and run scripts to quantize the model, instead, just run it in C#.

string srcFilename = "<Your source path>";
string dstFilename = "<Your destination path>";
string ftype = "q4_0";
if(Quantizer.Quantize(srcFileName, dstFilename, ftype))
{
    Console.WriteLine("Quantization succeed!");
}
else
{
    Console.WriteLine("Quantization failed!");
}

For more usages, please refer to Examples.

Web API

We provide the integration of ASP.NET core here. Since currently the API is not stable, please clone the repo and use it. In the future we'll publish it on NuGet.

Demo

demo-console

Roadmap

✅ LLaMa model inference.

✅ Embeddings generation.

✅ Chat session.

✅ Quantization

✅ ASP.NET core Integration

🔳 UI Integration

🔳 Follow up llama.cpp and improve performance

Assets

The model weights are too large to be included in the repository. However some resources could be found below:

The weights included in the magnet is exactly the weights from Facebook LLaMa.

The prompts could be found below:

Contact us

Join our chat on Discord.

License

This project is licensed under the terms of the MIT license.

Product Compatible and additional computed target framework versions.
.NET net5.0 was computed.  net5.0-windows was computed.  net6.0 is compatible.  net6.0-android was computed.  net6.0-ios was computed.  net6.0-maccatalyst was computed.  net6.0-macos was computed.  net6.0-tvos was computed.  net6.0-windows was computed.  net7.0 is compatible.  net7.0-android was computed.  net7.0-ios was computed.  net7.0-maccatalyst was computed.  net7.0-macos was computed.  net7.0-tvos was computed.  net7.0-windows was computed.  net8.0 was computed.  net8.0-android was computed.  net8.0-browser was computed.  net8.0-ios was computed.  net8.0-maccatalyst was computed.  net8.0-macos was computed.  net8.0-tvos was computed.  net8.0-windows was computed. 
.NET Core netcoreapp2.0 was computed.  netcoreapp2.1 was computed.  netcoreapp2.2 was computed.  netcoreapp3.0 was computed.  netcoreapp3.1 was computed. 
.NET Standard netstandard2.0 is compatible.  netstandard2.1 was computed. 
.NET Framework net461 was computed.  net462 was computed.  net463 was computed.  net47 was computed.  net471 was computed.  net472 was computed.  net48 was computed.  net481 was computed. 
MonoAndroid monoandroid was computed. 
MonoMac monomac was computed. 
MonoTouch monotouch was computed. 
Tizen tizen40 was computed.  tizen60 was computed. 
Xamarin.iOS xamarinios was computed. 
Xamarin.Mac xamarinmac was computed. 
Xamarin.TVOS xamarintvos was computed. 
Xamarin.WatchOS xamarinwatchos was computed. 
Compatible target framework(s)
Included target framework(s) (in package)
Learn more about Target Frameworks and .NET Standard.

NuGet packages (10)

Showing the top 5 NuGet packages that depend on LLamaSharp:

Package Downloads
Microsoft.KernelMemory.AI.LlamaSharp

Provide access to OpenAI LLM models in Kernel Memory to generate text
LangChain.Providers.LLamaSharp

LLamaSharp Chat model provider.
LLamaSharp.semantic-kernel

The integration of LLamaSharp and Microsoft semantic-kernel.
LLamaSharp.kernel-memory

The integration of LLamaSharp and Microsoft kernel-memory. It could make it easy to support document search for LLamaSharp model inference.
LangChain.Providers.Automatic1111

Automatic1111 Stable DIffusion model provider.

GitHub repositories (4)

Showing the top 4 popular GitHub repositories that depend on LLamaSharp:

Repository Stars
SciSharp/BotSharp
The AI Agent Framework in .NET
microsoft/kernel-memory
RAG architecture: index and query any data using LLM and natural language, track sources, show citations, asynchronous memory patterns.
CodeMazeBlog/CodeMazeGuides
The main repository for all the Code Maze guides
jxq1997216/AITranslator
使用大语言模型来翻译MTool导出的待翻译文件的图像化UI软件
Version Downloads Last updated
0.18.0 2,925 10/19/2024
0.17.0 673 10/13/2024
0.16.0 40,417 9/1/2024
0.15.0 16,635 8/3/2024
0.14.0 2,991 7/16/2024
0.13.0 47,602 6/4/2024
0.12.0 77,958 5/12/2024
0.11.2 14,423 4/6/2024
0.11.1 834 3/31/2024
0.10.0 6,189 2/15/2024
0.9.1 9,854 1/6/2024
0.9.0 543 1/6/2024
0.8.1 21,150 11/28/2023
0.8.0 19,676 11/12/2023
0.7.0 1,843 10/31/2023
0.6.0 2,366 10/24/2023
0.5.1 6,890 9/5/2023
0.4.2-preview 1,942 8/6/2023
0.4.1-preview 1,244 6/21/2023
0.4.0 11,343 6/19/2023
0.3.0 10,674 5/22/2023
0.2.3 711 5/17/2023
0.2.2 636 5/17/2023
0.2.1 673 5/12/2023
0.2.0 839 5/12/2023

LLama 0.2.3 mainly fixed some BUGs of model inference.