HyperparameterTrial<T>

Models & Types · AiDotNet.Models

Represents a single trial in hyperparameter optimization.

How It Works

For Beginners: A trial is one attempt at training with a specific set of hyperparameters. The optimizer tries many trials to find the best hyperparameter values.

Constructors

Constructor Summary
HyperparameterTrial(int trialNumber) Initializes a new instance of the HyperparameterTrial class.

Properties

Property Summary
DateTime StartTime { get; set; } Gets or sets the start time of the trial.
DateTime? EndTime { get; set; } Gets or sets the end time of the trial.
Dictionary<int, T> IntermediateValues { get; set; } Gets or sets intermediate values logged during training.
Dictionary<string, object> Parameters { get; set; } Gets or sets the hyperparameter values for this trial.
Dictionary<string, object> SystemAttributes { get; set; } Gets or sets system attributes for the trial.
Dictionary<string, object> UserAttributes { get; set; } Gets or sets user attributes for the trial.
T ObjectiveValue { get; set; } Gets or sets the objective value achieved (e.g., validation accuracy).
TrialStatus Status { get; set; } Gets or sets the status of the trial.
int TrialNumber { get; set; } Gets the trial number (sequential).
string TrialId { get; set; } Gets the unique identifier for this trial.

Methods

Method Summary
TimeSpan? GetDuration() Gets the duration of the trial.
void Complete(T objectiveValue) Marks the trial as complete with the final objective value.
void Fail() Marks the trial as failed.
void Prune() Marks the trial as pruned (stopped early).
void ReportIntermediateValue(int step, T value) Reports an intermediate value for a training step.