Class BenchmarkReport

java.lang.Object
ai.redouble.nucleo.tools.benchmark.BenchmarkReport
public class BenchmarkReport extends Object
What a benchmark measured: the job class and the grade it raced, the reference row, one row per model and run, and the means per model. Rendered as a table for the log by table(); a host persists the object where it keeps such things.
Since:
0.1 (2026-09-17)
Author:
Andrey Santrosyan

  • Constructor Details

    • BenchmarkReport

      public BenchmarkReport()

  • Method Details

    • getJobClass

      public String getJobClass()

    • setJobClass

      public void setJobClass(String jobClass)

    • getGrade

      public Grade getGrade()

    • setGrade

      public void setGrade(Grade grade)

    • getRuns

      public int getRuns()

    • setRuns

      public void setRuns(int runs)

    • getJudgeModel

      public String getJudgeModel()

    • setJudgeModel

      public void setJudgeModel(String judgeModel)

    • getJudgeFailure

      public String getJudgeFailure()

    • setJudgeFailure

      public void setJudgeFailure(String judgeFailure)

    • getElapsedMs

      public long getElapsedMs()

    • setElapsedMs

      public void setElapsedMs(long elapsedMs)

    • getRows

      public List<BenchmarkReport.Row> getRows()

    • setRows

      public void setRows(List<BenchmarkReport.Row> rows)

    • getSummaries

      public List<BenchmarkReport.Summary> getSummaries()

    • setSummaries

      public void setSummaries(List<BenchmarkReport.Summary> summaries)

    • table

      public String table()
      The means per model as a fixed-width table, the reference first, for a log line. Coloured for a terminal: failures green at zero, orange up to a quarter of the runs, red past that; cost per run heat-mapped from the cheapest model (green) to the dearest (red); the judge's score and the scorer's from their best (green) to their worst (red) on their own scales.