SafetyBenchmarkResult
Models & Types · AiDotNet.Safety.Benchmarking
Results from running a safety benchmark.
Constructors
| Constructor |
Summary |
SafetyBenchmarkResult() |
|
Properties
| Property |
Summary |
IReadOnlyDictionary<SafetyCategory, SafetyBenchmarkCategoryResult> CategoryResults { get; set; } |
Per-category benchmark results. |
double AverageLatencyMs { get; set; } |
Average evaluation latency in milliseconds. |
double F1Score { get; set; } |
F1 Score: Harmonic mean of precision and recall. |
double FalsePositiveRate { get; set; } |
False Positive Rate: FP / (FP + TN). |
double Precision { get; set; } |
Precision: TP / (TP + FP). |
double Recall { get; set; } |
Recall: TP / (TP + FN). |
int FalseNegatives { get; set; } |
Number of unsafe content missed. |
int FalsePositives { get; set; } |
Number of safe content incorrectly flagged. |
int TotalTestCases { get; set; } |
Total number of test cases evaluated. |
int TrueNegatives { get; set; } |
Number of correctly identified safe content. |
int TruePositives { get; set; } |
Number of correctly identified unsafe content. |
Fields
| Field |
Summary |
SafetyBenchmarkResult Empty |
An empty benchmark result. |