RLEpisodeMetrics<T>

Models & Types · AiDotNet.Configuration

Metrics for a completed RL episode.

Constructors

Constructor Summary
RLEpisodeMetrics() Creates a new instance with default values.

Properties

Property Summary
T AverageLoss { get; set; } Average loss during training in this episode.
T AverageRewardRecent { get; set; } Running average reward over recent episodes (smoothed metric).
T TotalReward { get; set; } Total reward accumulated in this episode.
TimeSpan ElapsedTime { get; set; } Total time elapsed since training started.
bool TerminatedNaturally { get; set; } Whether the episode ended naturally (vs hitting max steps).
int Episode { get; set; } The episode number (1-indexed).
int Steps { get; set; } Number of steps taken in this episode.