Reports
Every completed validation run produces a report with scores across four dimensions, detailed findings in JSON, and optional annotated images highlighting detected issues.Scoring
Reports score your dataset from 0 to 100 on each dimension, plus an overall composite score.Overall
Weighted composite of all dimension scores.
Coverage
Scenario and class coverage completeness.
Physics
Physical plausibility of the synthetic data.
Distribution
Class balance and spatial distribution quality.
Sim-to-Real
Domain gap relative to real-world reference data.
Get a Report
Retrieve the full report for a completed validation run.Response
Response
Unique report identifier.
The validation run that produced this report.
The dataset that was validated.
The project this report belongs to.
Composite score from 0 to 100.
Coverage dimension score from 0 to 100.
Physics plausibility score from 0 to 100.
Distribution quality score from 0 to 100.
Sim-to-real gap score from 0 to 100. Null if no reference dataset was provided.
Detailed findings for each validation dimension.
GCS path where report artifacts are stored.
List of image keys available for download via the images endpoint.
ISO 8601 timestamp of report generation.
Export as PDF
Download a formatted PDF version of the report suitable for sharing with stakeholders or attaching to compliance documentation.The PDF endpoint returns a binary stream with
Content-Type: application/pdf. Use the -o flag in cURL or write the response content as bytes in your client.Annotated Images
Retrieve annotated images that highlight specific issues found during validation. The endpoint returns a 307 redirect to a time-limited signed GCS URL.Score Trends
Track how validation scores change over time across multiple runs within a project. Use this to monitor data quality regressions or improvements across retraining cycles.The score dimension to query. One of
overall, coverage, physics, distribution, or sim_to_real.Start date in
YYYY-MM-DD format.End date in
YYYY-MM-DD format.Response
Response
Ordered list of score data points within the requested time range.
Date of the validation run.
Score value for the requested metric (0-100).
The validation run that produced this data point.
Trends are available for all five metrics:
overall, coverage, physics, distribution, and sim_to_real. Query each separately or build a multi-metric dashboard by making parallel requests.