From 2fd35237775336caa374dc5b61d4edf9fd0e56b7 Mon Sep 17 00:00:00 2001 From: "russell@unturf.com" Date: Thu, 21 May 2026 12:37:24 -0400 Subject: [PATCH] fix(#000057): control_ab header prints actual model/answer_mode/judge MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Was a stale hardcoded 'same Hermes; judge=Opus hermetic' label that misreported any run with --model/--judge overrides (e.g. qwen + code judge). Now reflects the real config — honest header for the artifact. --- bench/control_ab.py | 5 +++-- 1 file changed, 3 insertions(+), 2 deletions(-) diff --git a/bench/control_ab.py b/bench/control_ab.py index 7a4988c..fb851ed 100644 --- a/bench/control_ab.py +++ b/bench/control_ab.py @@ -219,8 +219,9 @@ def main() -> int: def rate(arm, *labels): return sum(tally[arm][x] for x in labels) - print(f"\n=== #000057 v1 control A/B (n={n}, single-shot, same " - f"Hermes; judge=Opus hermetic) ===") + print(f"\n=== #000057 v1 control A/B (n={n}, single-shot, " + f"model={a.model}; answer_mode={a.answer_mode}; " + f"judge={a.judge}) ===") print(f"fixture={a.fixture} log={outp}") for arm in ("solo", "arborist"): c = tally[arm]