Table of Contents

Class RedTeamingResults<T>

Namespace
AiDotNet.Models
Assembly
AiDotNet.dll

Contains results from red teaming adversarial testing.

public class RedTeamingResults<T>

Type Parameters

T

The numeric data type used for calculations.

Inheritance
RedTeamingResults<T>
Inherited Members

Properties

AdversarialPrompts

Gets or sets the adversarial prompts that were tested.

public Matrix<T> AdversarialPrompts { get; set; }

Property Value

Matrix<T>

AverageSeverity

Gets or sets the average severity of successful attacks.

public double AverageSeverity { get; set; }

Property Value

double

ModelResponses

Gets or sets the model's responses to adversarial prompts.

public Matrix<T> ModelResponses { get; set; }

Property Value

Matrix<T>

SeverityScores

Gets or sets the severity scores for each vulnerability (0-1).

public double[] SeverityScores { get; set; }

Property Value

double[]

SuccessRate

Gets or sets the overall red teaming success rate.

public double SuccessRate { get; set; }

Property Value

double

SuccessfulAttacks

Gets or sets which prompts successfully caused misaligned behavior.

public bool[] SuccessfulAttacks { get; set; }

Property Value

bool[]

Vulnerabilities

Gets or sets detailed descriptions of vulnerabilities found.

public List<VulnerabilityReport> Vulnerabilities { get; set; }

Property Value

List<VulnerabilityReport>

VulnerabilityTypes

Gets or sets the types of vulnerabilities found.

public string[] VulnerabilityTypes { get; set; }

Property Value

string[]