之前的篇章都是用SemanticKernel来连接OpenAI的API,当然是需要费用,另外还有使用限制,本篇来说明在SK中使用开源模型LLama3。
首先引入Nuget包,这里使用的是LLamaSharp这个三方包,因为没有显卡,只能跑在CPU上,所以也需要引入对应的Cpu包,最后引入SK的LLama版的包。
<ItemGroup>
<PackageReference Include="LLamaSharp" Version="0.11.2" />
<PackageReference Include="LLamaSharp.Backend.Cpu" Version="0.11.2" />
<PackageReference Include="LLamaSharp.semantic-kernel" Version="0.11.2" />
</ItemGroup>
接下就是下载最新的LLama3了,扩展名是gguf,如下代码就可以轻松地跑起本地小模型了。
using LLama.Common;
using LLama;
using LLamaSharp.SemanticKernel.ChatCompletion;
using System.Text;
using ChatHistory = LLama.Common.ChatHistory;
using AuthorRole = LLama.Common.AuthorRole;
await SKRunAsync();
async Task SKRunAsync()
{
var modelPath = @"C:\llama\llama-2-coder-7b.Q8_0.gguf";
var parameters = new ModelParams(modelPath)
{
ContextSize = 1024,
Seed = 1337,
GpuLayerCount = 5,
Encoding = Encoding.UTF8,
};
using var model = LLamaWeights.LoadFromFile(parameters);
var ex = new StatelessExecutor(model, parameters);
var chatGPT = new LLamaSharpChatCompletion(ex);
var chatHistory = chatGPT.CreateNewChat(@"这是assistant和user之间的对话。
assistant是一名.net和C#专家,能准确回答user提出的专业问题。");
Console.WriteLine("开始聊天:");
Console.WriteLine("------------------------");
while (true)
{
Console.Write("user:");
var userMessage = Console.ReadLine();
chatHistory.AddUserMessage(userMessage);
var first = true;
var content = "";
await foreach (var reply in chatGPT.GetStreamingChatMessageContentsAsync(chatHistory))
{
if (first)
{
first = false;
Console.Write(reply.Role + ":");
}
content += reply.Content;
Console.Write(reply.Content);
}
chatHistory.AddAssistantMessage(content);
}
}
下面是具体的效果,除了慢点,没有GPT强大点,其他都是很香的,关键是没有key,轻松跑,不怕信用卡超支。