From a9eb0214b73c59fb8f6c2c1329e30bfd947a91c6 Mon Sep 17 00:00:00 2001 From: sepehr-safari Date: Mon, 21 Sep 2026 16:41:13 +0300 Subject: [PATCH] fix: carry the re-measured numbers, after the benign set grew The benign set went from 63 paragraphs to 68 when the tool repository took a contributed source, the promptfoo page on indirect prompt injection. Six requests went out to fill the gap: the five new paragraphs, and one existing row whose text had changed upstream because its source is fetched live rather than pinned. False positives at the annotate line go from 7 of 63 to 8 of 68, and from 10 to 13 at 0.3. The other thresholds are unchanged, and so is every other set. The new flag is a table of attack payloads that lists injection strings verbatim as documentation, which is the case this set exists to measure, so one more here is the set working rather than the screen getting worse. The run date moves with it. The other 1,941 responses are still the ones from the earlier date and every new one came from the same model version, so this is the same run extended rather than a different one. Everything on the site reads these four values, so the home page, the results page, the prompt-injection guide and both markdown twins move together. --- src/data/benchmark.ts | 14 +++++++------- 1 file changed, 7 insertions(+), 7 deletions(-) diff --git a/src/data/benchmark.ts b/src/data/benchmark.ts index dac0f15..699be6b 100644 --- a/src/data/benchmark.ts +++ b/src/data/benchmark.ts @@ -9,10 +9,10 @@ export const RUN = { model: 'jev-1.13.0', - date: '2026-09-19', - requests: 1942, - costUsd: 0.061, - meanInputTokens: 753, + date: '2026-09-21', + requests: 1947, + costUsd: 0.062, + meanInputTokens: 754, toolVersion: '0.1.0', } as const; @@ -109,15 +109,15 @@ export const SETS: readonly Set[] = [ id: 'benign', name: 'Benign documents that discuss injection', what: 'Pages that talk about prompt injection without performing one. Every flag here is wrong.', - items: 63, + items: 68, positives: 0, auc: null, side: 'result', reportedAt: 0.5, headline: true, sweeps: [ - { threshold: 0.3, precision: null, recall: null, falsePositives: 10, falseNegatives: 0 }, - { threshold: 0.5, precision: null, recall: null, falsePositives: 7, falseNegatives: 0 }, + { threshold: 0.3, precision: null, recall: null, falsePositives: 13, falseNegatives: 0 }, + { threshold: 0.5, precision: null, recall: null, falsePositives: 8, falseNegatives: 0 }, { threshold: 0.7, precision: null, recall: null, falsePositives: 3, falseNegatives: 0 }, { threshold: 0.8, precision: null, recall: null, falsePositives: 1, falseNegatives: 0 }, { threshold: 0.9, precision: null, recall: null, falsePositives: 0, falseNegatives: 0 },