AlignmentFeedbackData<T>
Models & Types · AiDotNet.Models
Contains human feedback data for AI alignment.
Constructors
| Constructor |
Summary |
AlignmentFeedbackData() |
|
Properties
| Property |
Summary |
Matrix<T> Inputs { get; set; } |
Gets or sets the input prompts or examples. |
Matrix<T> Outputs { get; set; } |
Gets or sets the model outputs for each input. |
ValueTuple<int, int>[] Preferences { get; set; } |
Gets or sets human preference comparisons. |
double[] Ratings { get; set; } |
Gets or sets numerical ratings for each output (optional). |
double[] Rewards { get; set; } |
Gets or sets reward labels for reinforcement learning. |
string[] TextualFeedback { get; set; } |
Gets or sets textual feedback for outputs (optional). |
Methods
| Method |
Summary |
bool ValidatePreferences() |
Validates that preference indices are within valid bounds. |
void EnsurePreferencesValid() |
Validates preference indices and throws if invalid. |