Class AgentConfiguration<T>

Namespace

AiDotNet.Models

Assembly

AiDotNet.dll

Stores the configuration settings for AI agent assistance during model building and inference.

public class AgentConfiguration<T>

Type Parameters

T

The numeric type used for model parameters and calculations (typically float or double).

Inheritance

object

AgentConfiguration<T>

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

This class encapsulates all the settings needed to enable and configure AI agent assistance in the AiModelBuilder. It stores the API credentials, provider selection, and Azure-specific configuration required to connect to large language model (LLM) services. The configuration is used both during model building (for recommendations and analysis) and during inference (for conversational assistance with trained models).

For Beginners: This class holds the settings that tell the AI agent how to connect to language model services.

Think of it as a configuration card that contains:

• API Key: Your password/credentials for the AI service
• Provider: Which AI company to use (OpenAI, Anthropic, or Azure)
• IsEnabled: Whether agent assistance is turned on or off
• Azure Settings: Extra information needed if using Microsoft Azure

Why you need this:

• The AI agent needs credentials to call external AI services
• Different providers need different configuration
• This keeps all settings in one organized place
• The same configuration can be reused across multiple model builds

For example, if you're using OpenAI's GPT-4, this class would store your OpenAI API key and indicate that the provider is "OpenAI". If you're using Azure, it would also store your Azure endpoint URL and deployment name.

Security note: API keys are sensitive! This class is marked with [JsonIgnore] attributes to prevent accidentally saving keys to disk when saving your trained models.

Properties

ApiKey

Gets or sets the API key for authenticating with the LLM provider.

[JsonIgnore]
public string? ApiKey { get; set; }

Property Value

string

The API key string, or null if using alternative authentication methods like environment variables.

Remarks

The API key is used to authenticate requests to the LLM provider's API. If not provided explicitly, the system will attempt to resolve the key from global configuration or environment variables. For security, this value should not be hard-coded but loaded from secure configuration stores.

For Beginners: This is like a password that proves you're authorized to use the AI service.

Where to get API keys:

• OpenAI: Sign up at platform.openai.com and generate a key in your account settings
• Anthropic: Get a key from console.anthropic.com after creating an account
• Azure OpenAI: Create a key in the Azure Portal after getting approval for Azure OpenAI Service

You can provide the key in three ways:

1. Set it here directly (not recommended for production)
2. Set it globally at application startup using AgentGlobalConfiguration
3. Store it in an environment variable (OPENAI_API_KEY, ANTHROPIC_API_KEY, etc.)

Security tip: Never commit API keys to source control! Use environment variables or secret management services in production applications.

AssistanceOptions

Gets or sets the agent assistance options that control which types of help the agent provides.

public AgentAssistanceOptions? AssistanceOptions { get; set; }

Property Value

AgentAssistanceOptions

Agent assistance options, or null to use AgentAssistanceOptions.Default.

Remarks

For Beginners: This controls what the agent helps with during model building.

You can customize exactly which assistance features you want:

• Data analysis (examines your dataset characteristics)
• Model selection (recommends which model type to use)
• Hyperparameter tuning (suggests optimal parameter values)
• Feature analysis (identifies important features)
• Meta-learning advice (guidance on validation and regularization)

If you don't set this, it defaults to AgentAssistanceOptions.Default which enables data analysis and model selection (the most commonly useful features).

For full assistance, use AgentAssistanceOptions.Comprehensive. For minimal assistance, use AgentAssistanceOptions.Minimal.

AzureDeployment

Gets or sets the Azure OpenAI deployment name (required only when Provider is AzureOpenAI).

public string? AzureDeployment { get; set; }

Property Value

string

The name of the deployed model in your Azure OpenAI resource, or null if not using Azure OpenAI.

Remarks

Azure OpenAI requires you to deploy specific models with custom names. This property specifies which deployed model to use for agent operations. Common deployment names include variations of "gpt-4", "gpt-35-turbo", etc., but the exact name depends on what you named it when deploying.

For Beginners: When using Azure OpenAI, you must first "deploy" a model in the Azure Portal, giving it a name of your choice. That name goes here.

How to find your deployment name:

1. Go to Azure OpenAI Studio (oai.azure.com)
2. Navigate to "Deployments" in the left menu
3. You'll see a list of your deployed models with their names
4. Copy the name of the deployment you want to use

For example, if you deployed GPT-4 and named it "my-gpt4-model", you would set this property to "my-gpt4-model". This is different from regular OpenAI where you just specify the model directly.

AzureEndpoint

Gets or sets the Azure OpenAI endpoint URL (required only when Provider is AzureOpenAI).

public string? AzureEndpoint { get; set; }

Property Value

string

The full Azure OpenAI resource endpoint URL, or null if not using Azure OpenAI.

Remarks

This property is only used when the Provider is set to AzureOpenAI. The endpoint URL is specific to your Azure OpenAI resource and typically follows the format: https://YOUR-RESOURCE-NAME.openai.azure.com/

For Beginners: If you're using Azure OpenAI instead of regular OpenAI, you need to provide your Azure endpoint URL here.

How to find your Azure endpoint:

1. Go to the Azure Portal (portal.azure.com)
2. Navigate to your Azure OpenAI resource
3. Look for "Keys and Endpoint" in the left menu
4. Copy the endpoint URL (looks like https://your-name.openai.azure.com/)

This tells the agent where to send requests in Azure's cloud infrastructure instead of OpenAI's servers.

IsEnabled

Gets or sets a value indicating whether agent assistance is currently enabled.

public bool IsEnabled { get; set; }

Property Value

bool

True if the agent should provide assistance; false to disable agent features.

Remarks

For Beginners: This is an on/off switch for AI agent assistance.

When enabled (true):

• The agent can recommend model types
• The agent can suggest hyperparameters
• The agent can analyze your data characteristics
• You can ask questions during model building and inference

When disabled (false):

• All agent features are turned off
• No API calls are made to LLM services
• Model building proceeds without AI assistance

This is useful for temporarily disabling agent features without removing all configuration, or for testing your code with and without agent assistance.

Provider

Gets or sets the LLM provider to use for agent operations.

public LLMProvider Provider { get; set; }

Property Value

LLMProvider

An LLMProvider enum value indicating which service to use. Defaults to OpenAI.

Remarks

The provider determines which API endpoints are called and what authentication method is used. Different providers offer different model capabilities, pricing structures, and compliance features.