AlignmentMethodOptions<T>
Options & Configuration ยท AiDotNet.Models.Options
Configuration options for AI alignment methods.
Inheritance: ModelOptions โ AlignmentMethodOptions<T>
For Beginners
These settings control how your AI learns to behave according to human values. You can adjust how much to weight human feedback, what principles to follow, and how thoroughly to test for problems.
How It Works
These options control how models are aligned with human values and intentions through techniques like RLHF, constitutional AI, and red teaming.
Constructors
| Constructor | Summary |
|---|---|
AlignmentMethodOptions() |
Properties
| Property | Summary |
|---|---|
RedTeamingCategory[] RedTeamingCategories { get; set; } |
Gets or sets the safety-benchmark categories red teaming draws its prompts from. |
bool EnableRedTeaming { get; set; } |
Gets or sets whether to perform red teaming. |
bool UseConstitutionalAI { get; set; } |
Gets or sets whether to use constitutional AI principles. |
double Gamma { get; set; } |
Gets or sets the discount factor for reward modeling. |
double KLCoefficient { get; set; } |
Gets or sets the KL divergence penalty coefficient. |
double LearningRate { get; set; } |
Gets or sets the learning rate for alignment training. |
double RewardModelLearningRate { get; set; } |
Gets or sets the gradient-ascent step size of the Bradley-Terry reward model fit. |
int CritiqueIterations { get; set; } |
Gets or sets the number of critique iterations for constitutional AI. |
int RedTeamingAttempts { get; set; } |
Gets or sets the number of red teaming attempts. |
int TrainingIterations { get; set; } |
Gets or sets the number of training iterations for RLHF. |
string RewardModelArchitecture { get; set; } |
Gets or sets the reward model architecture. |