From 4865d62a86f62ffaa073a734a3ab3cbb69f1e8b8 Mon Sep 17 00:00:00 2001 From: YurenHao0426 Date: Tue, 21 Jul 2026 10:14:50 -0500 Subject: experiments: add fail-fast claim finalization --- experiments/finalize_claims.sh | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+) create mode 100755 experiments/finalize_claims.sh diff --git a/experiments/finalize_claims.sh b/experiments/finalize_claims.sh new file mode 100755 index 0000000..70e53a2 --- /dev/null +++ b/experiments/finalize_claims.sh @@ -0,0 +1,30 @@ +#!/usr/bin/env bash +# Fail-fast finalization for the two publication-facing claims. This command +# intentionally fails until every strict figure cell has clean seeds 0..4. +set -eu + +cd "$(dirname "$0")/.." +TRAIN_PY=/home/yurenh2/miniconda3/envs/ep_pascal/bin/python3 +PLOT_PY=/home/yurenh2/miniconda3/bin/python +OUTDIR=results/figs +SUMMARY=results/audited_tables.md + +mkdir -p "$OUTDIR" + +# Run the strongest gate first. It validates exact seed sets, provenance, +# protocols, finite metrics, and only then writes figures + hash manifest. +"$PLOT_PY" experiments/plot_main_figures.py \ + --results results --outdir "$OUTDIR" + +"$TRAIN_PY" experiments/baseline_smoke.py +"$TRAIN_PY" experiments/synthetic_smoke.py +"$TRAIN_PY" experiments/smoke.py + +summary_tmp=$(mktemp /tmp/sdil_audited_tables.XXXXXX) +trap 'rm -f "$summary_tmp"' EXIT +"$TRAIN_PY" experiments/analyze_verified.py --results results > "$summary_tmp" +mv "$summary_tmp" "$SUMMARY" +trap - EXIT + +git diff --check +echo "finalized audited claims -> $OUTDIR and $SUMMARY" -- cgit v1.2.3