Experiment log

Experiment log

For each run retain:

  • Mission, target version, model/tool identification where available.
  • Exact input and initial output file.
  • First invalid inference, or reason the output passes.
  • Repair requested and resulting evidence.
  • Fresh transfer case and its observed outcome.
  • Time/resource observations if actually measured.
  • What the mechanism improved; what remains untested.

Do not fill this log with predicted or invented results.