Semantic Kernel：接入azure中的deepseek-r1-编程知识

Semantic Kernel：接入azure中的deepseek-r1

news/2025/3/3 21:41:07/文章来源:https://www.cnblogs.com/axzxs2001/p/18746273

　　SemanticKernel已经支持deepseek-r1了，官方的Blog地址是https://devblogs.microsoft.com/semantic-kernel/using-deepseek-models-in-semantic-kernel，同时给出了接入的Demo，遗憾的是deepseek不支持API的充值，没有办法测试。

　　办法总比困难多，正好微软现在支持在Azure部署deepseek-r1了，具体部署方法官方有具体的方式方法，见下面的url：https://azure.microsoft.com/en-us/blog/deepseek-r1-is-now-available-on-azure-ai-foundry-and-github/

　　与此同时，Azure上的给Demo如下：

var apiKey = File.ReadAllText("C:/gpt/deepseekkey.txt");
var handler = new HttpClientHandler(){ClientCertificateOptions = ClientCertificateOption.Manual,ServerCertificateCustomValidationCallback =(httpRequestMessage, cert, cetChain, policyErrors) => { return true; }};
using (var client = new HttpClient(handler))
{var requestBody = @"{""messages"": [                 {""role"": ""user"",""content"": ""你是谁？""}],""max_tokens"": 2048}";client.DefaultRequestHeaders.Authorization = new AuthenticationHeaderValue("Bearer", apiKey);client.BaseAddress = new Uri("https://DeepSeek-R1-iwztj.eastus2.models.ai.azure.com/chat/completions");var content = new StringContent(requestBody);content.Headers.ContentType = new MediaTypeHeaderValue("application/json");HttpResponseMessage response = await client.PostAsync("", content);if (response.IsSuccessStatusCode){string result = await response.Content.ReadAsStringAsync();var ent = System.Text.Json.JsonSerializer.Deserialize<ChatCompletionResponse>(result, new System.Text.Json.JsonSerializerOptions { PropertyNameCaseInsensitive = true });foreach (var choice in ent.Choices){Console.WriteLine("Result: {0}", choice.Message.Content);}}else{Console.WriteLine(string.Format("The request failed with status code: {0}", response.StatusCode));Console.WriteLine(response.Headers.ToString());Console.WriteLine("错误");}
}public class ChatCompletionResponse
{public List<Choice> Choices { get; set; }public long Created { get; set; }public string Id { get; set; }public string Model { get; set; }public string Object { get; set; }public List<PromptFilterResult> PromptFilterResults { get; set; }public Usage Usage { get; set; }
}public class Choice
{public ContentFilterResults ContentFilterResults { get; set; }public string FinishReason { get; set; }public int Index { get; set; }public Message Message { get; set; }
}public class ContentFilterResults
{public FilterStatus Hate { get; set; }public FilterStatus SelfHarm { get; set; }public FilterStatus Sexual { get; set; }public FilterStatus Violence { get; set; }
}public class FilterStatus
{public bool Filtered { get; set; }public string Severity { get; set; }
}public class Message
{public string Content { get; set; }public string ReasoningContent { get; set; }public string Role { get; set; }public object ToolCalls { get; set; }
}public class PromptFilterResult
{public int PromptIndex { get; set; }public ContentFilterResults ContentFilterResults { get; set; }
}public class Usage
{public int CompletionTokens { get; set; }public int PromptTokens { get; set; }public object PromptTokensDetails { get; set; }public int TotalTokens { get; set; }
}

结果如下：

换成SK的对接方式如下：

var apiKey = File.ReadAllText("C:/gpt/deepseekkey.txt");
var chatCompletionService = new OpenAIChatCompletionService(endpoint: new Uri("https://DeepSeek-R1-iwztj.eastus2.models.ai.azure.com/"),apiKey: apiKey,modelId: "deepseek-chat"
);
var chatHistory = new ChatHistory();
chatHistory.AddUserMessage("你好，你是谁?");
var reply = await chatCompletionService.GetChatMessageContentAsync(chatHistory);
Console.WriteLine(reply);