Try Gemini 1.5 Pro, our most advanced multimodal model in Vertex AI, and see what you can build with a 1M token context window. Try Gemini 1.5 Pro, our most advanced multimodal model in Vertex AI, and see what you can build with a 1M token context window.

MultiConfidenceMetrics

Metrics across multiple confidence levels.

JSON representation

JSON representation
{ "confidenceLevelMetrics": [ { object (`ConfidenceLevelMetrics`) } ], "confidenceLevelMetricsExact": [ { object (`ConfidenceLevelMetrics`) } ], "auprc": number, "estimatedCalibrationError": number, "auprcExact": number, "estimatedCalibrationErrorExact": number, "metricsType": enum (`MetricsType`) }

{
  "confidenceLevelMetrics": [
    {
      object (ConfidenceLevelMetrics)
    }
  ],
  "confidenceLevelMetricsExact": [
    {
      object (ConfidenceLevelMetrics)
    }
  ],
  "auprc": number,
  "estimatedCalibrationError": number,
  "auprcExact": number,
  "estimatedCalibrationErrorExact": number,
  "metricsType": enum (MetricsType)
}

Fields
`confidenceLevelMetrics[]`	`object (ConfidenceLevelMetrics)` Metrics across confidence levels with fuzzy matching enabled.
`confidenceLevelMetricsExact[]`	`object (ConfidenceLevelMetrics)` Metrics across confidence levels with only exact matching.
`auprc`	`number` The calculated area under the precision recall curve (AUPRC), computed by integrating over all confidence thresholds.
`estimatedCalibrationError`	`number` The Estimated Calibration Error (ECE) of the confidence of the predicted entities.
`auprcExact`	`number` The AUPRC for metrics with fuzzy matching disabled, i.e., exact matching only.
`estimatedCalibrationErrorExact`	`number` The ECE for the predicted entities with fuzzy matching disabled, i.e., exact matching only.
`metricsType`	`enum (MetricsType)` The metrics type for the label.