Here's an extension method to use a chat completion service as a text completion service

[guillaume86]

I wanted to use the cheaper gpt3.5-turbo, here's the revelant code:

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Threading.Tasks;
using Microsoft.SemanticKernel.AI.ChatCompletion;
using Microsoft.SemanticKernel.AI.TextCompletion;

namespace Microsoft.SemanticKernel;

public static class KernelConfigExtensions
{
    public static KernelConfig AddTextCompletionChatAdapter(this KernelConfig config, string serviceId, string chatServiceId = "")
    {
        return config.AddTextCompletionService(
            serviceId,
            (kernel) =>
            {
                var chatService = kernel.GetService<IChatCompletion>(chatServiceId);
                return new TextCompletionChatAdapter(chatService);
            });
    }
}

internal class TextCompletionChatAdapter : ITextCompletion
{
    public TextCompletionChatAdapter(IChatCompletion chatService)
    {
        this.ChatService = chatService;
    }

    public IChatCompletion ChatService { get; }

    public async Task<string> CompleteAsync(string text, CompleteRequestSettings requestSettings, CancellationToken cancellationToken = default)
    {
        var chat = this.ChatService.CreateNewChat(text);

        var message = await this.ChatService.GenerateMessageAsync(chat, new ChatRequestSettings
        {
            FrequencyPenalty = requestSettings.FrequencyPenalty,
            MaxTokens = requestSettings.MaxTokens,
            PresencePenalty = requestSettings.PresencePenalty,
            StopSequences = requestSettings.StopSequences,
            Temperature = requestSettings.Temperature,
            TopP = requestSettings.TopP,
        }, cancellationToken);

        return message;
    }
}

[dluc]

hi @guillaume86 this feature is now officially supported in the latest version 0.12 we just released :-)
see for example here:

IKernel kernel = new KernelBuilder().WithLogger(ConsoleLogger.Log).Build();

// Note: we use Chat Completion and GPT 3.5 Turbo
kernel.Config
    .AddAzureChatCompletionService("id", "gpt-35-turbo", "https://....openai.azure.com/", "...API KEY...");

var func = kernel.CreateSemanticFunction(
    "List the two planets closest to '{{$input}}', excluding moons, using bullet points.");

var result = await func.InvokeAsync("Jupiter");
Console.WriteLine(result);

[precisioninfinity]

example link is broken, here's the corrected one

@dluc, thanks for posting. Where's the documentation on if both chat and text services are registered which one is used? And also any examples of how you send the extra context information available by chat, namely you can send system, then chain of user, assistant history.