All evidence dossiers
FIDE AI / EVIDENCE DOSSIERMethod change · Inspect Evals contributors

When the score changed with the denominator

Before interpreting score movement as a model change, check whether the grader and denominator still count the same samples.

4 observations1 cited sourcePractice + inspection
HOW TO READ THIS RECORD

Make a call as evidence arrives, or inspect the complete sequence.

THE EVIDENCE LEDGER

What changes as the record unfolds?

These observations are curated paraphrases. Each line says what that part of the cited record can support.

01
TASK / PUBLISHED METHOD

A judge grades a prompt-injection response.

Inspect’s CyberSecEval 4 adaptation includes a multilingual prompt-injection task whose grader expects a parseable verdict.

SUPPORTSThe scorer is part of the measurement, not merely a display detail.
TRACE THIS OBSERVATION
Inspect Evals contributors

Scoring: Multilingual Prompt Injection

Original
02
EARLIER RULE / DOCUMENTED PRIOR BEHAVIOR

Unparseable grades counted as zero.

The published changelog says that earlier scoring treated a judge response without a parseable verdict as a 0.0 result.

SUPPORTSAn instrument failure was previously mixed into the scored denominator.
TRACE THIS OBSERVATION
Inspect Evals contributors

Changelog [4-B], August 20, 2026

Original
03
REVISED RULE / DOCUMENTED SCORING CHANGE

The same grades became unscored.

Version 4-B excludes those samples from accuracy and epoch-mean denominators and reports them as unscored.

SUPPORTSA score can change when the set of scored samples changes.
TRACE THIS OBSERVATION
Inspect Evals contributors

Changelog [4-B], August 20, 2026

Original
04
LIMIT / PUBLISHED COMPARABILITY WARNING

Scores across versions are not directly comparable.

The implementation warns that this task’s version 4-B results are not comparable with 3-B on that scoring basis.

SUPPORTSReport version and unscored counts before attributing a difference to the model.
TRACE THIS OBSERVATION
Inspect Evals contributors

Changelog [4-B]: comparability with 3-B

Original
THE OPEN RESEARCH TRAIL

What would we need to know next?