PredictionGuard.NET Class Library

PredictionGuard.NET is a .NET Standard 2.1 wrapper for the PredictionGuard API. The library supports the chat completions endpoint (with and without streaming), including function calling.

Features

• Chat Completions: Generate text completions based on input prompts.
• Function Calling: Invoke functions based on chat input.
• Embeddings: Generate embeddings for text and image inputs.

Usage

Chat Completions

Add the PredictionGuard Chat Client to the service container

builder.Services.AddPredictionGuardChatClient(config["ApiKey"]);

Add the PredictionGuard Chat Client with Options

Customize the endpoint, specify a model, etc.

builder.Services.AddPredictionGuardChatClient(config["ApiKey"], options =>
{
    options.Endpoint = new Uri(config["Endpoint"]);
    options.Model = config["Model"];
});

Retrieve the PredictionGuard Chat Client from the Service Container through dependency injection

Console Application

var predictionGuardChatClient = app.Services.GetRequiredService<PredictionGuardChatClient>();

Blazor

@inject PredictionGuardChatClient PredictionGuardChatClient

API controller

private readonly PredictionGuardChatClient _predictionGuardChatClient;

public MyController(PredictionGuardChatClient predictionGuardChatClient)
{
    _predictionGuardChatClient = predictionGuardChatClient;
}

Generate Chat Completions

var responseText = await predictionGuardChatClient.CompleteAsync(input);

Generate Chat Completions with Streaming

await foreach (var chunk in predictionGuardChatClient.CompleteStreamingAsync(input))
{
    Console.Write(chunk);
}

Enable Function Calling and Make Tools Available

The library is capable of invoking one or more functions to assist with the user query. Function output is sent back to the API, which then generates a natural language response based on the data.

builder.Services.AddPredictionGuardChatClient(config["ApiKey"])
    .UseFunctionInvocation();

MethodInfo getForecastMethod = ToolBuilder.GetMethod<WeatherService>("GetForecast");
ChatOptions chatOptions = new()
{
    Tools = [ getForecastMethod ]
};

var predictionGuardChatClient = app.Services.GetRequiredService<PredictionGuardChatClient>();

var responseText = await predictionGuardChatClient.CompleteAsync("Do I need an umbrella today in Nantes?", chatOptions);

public class Weather
{
    public int Day { get; set; }
    public string Location { get; set; }    
    public string Summary { get; set; }    
}

[Description("Gets a weather forecast for the given number of days")]
public List<Weather> GetForecast(string location, int numDays)
{
    List<Weather> forecast = new();
    for (int i = 0; i < numDays; i++)
    {
        forecast.Add(new Weather() { Day = i, Location = location, Summary = Random.Shared.NextDouble() > 0.5 ? "Sunny" : "Rainy" });
    }
    return forecast;
}

Embeddings

Add the PredictionGuard Embeddings Client to the service container

builder.Services.AddPredictionGuardEmbeddingsClient(config["ApiKey"]);

Add the PredictionGuard Embeddings Client with Options

Customize the endpoint, specify a model, etc.

builder.Services.AddPredictionGuardEmbeddingsClient(config["ApiKey"], options =>
{
    options.Endpoint = new Uri(config["Endpoint"]);
    options.Model = config["Model"];
});

Generate Embeddings from a Single Text Input

var embeddings = await predictionGuardEmbeddingsClient.GenerateEmbeddingsAsync("Your input text here");

Generate Embeddings from Multiple Inputs

var inputs = new List<EmbeddingInput>
{
    new EmbeddingInput
    {
        Text = "Your input text here"
    },
    new EmbeddingInput
    {
        Text = "Another input"
    }
};
var embeddings = await predictionGuardEmbeddingsClient.GenerateEmbeddingsAsync(inputs);
    foreach (var embeddingData in embeddings)
    {
        Console.WriteLine($"Index: {embeddingData.Index}");
        Console.WriteLine($"Embedding: {string.Join(", ", embeddingData.Embedding)}");
    }