From 33b7b7724dccfde7ffc424e023dcd81eb35114c6 Mon Sep 17 00:00:00 2001 From: byte271 Date: Tue, 25 Aug 2026 15:41:11 -0400 Subject: [PATCH 1/3] Start Cycle 7 with a collapse-resistant transform channel and Stage A evidence. Cycle 6 remains frozen at NONZERO_RESIDUAL 7/192. Fresh seed 810000 shows Cycle 6 spacing dies under whitespace collapse, while the first durable family lacks density on GPT-2 TinyDev text. Co-authored-by: Cursor --- docs/cycle7/README.md | 70 ++++ docs/cycle7/literature.md | 24 ++ docs/cycle7/live-state-audit.md | 57 +++ docs/cycle7/seed-ledger.md | 26 ++ docs/cycle7/stage-a-decision.md | 52 +++ docs/cycle7/whitespace-collapse-v1.md | 27 ++ evidence/cycle7-stage-a-2026-08-25/README.md | 25 ++ .../cycle7-stage-a-2026-08-25/SHA256SUMS.txt | 6 + .../decision-geometry-only.json | 1 + .../cycle7-stage-a-2026-08-25/decision.json | 1 + .../detector-compare-family1-pre-copula.json | 1 + .../detector-compare.json | 1 + .../fixture-geometry.json | 1 + fuckmark/cycle7/__init__.py | 37 ++ fuckmark/cycle7/compare.py | 103 +++++ fuckmark/cycle7/decision.py | 137 ++++++ fuckmark/cycle7/durable_rules.py | 337 +++++++++++++++ fuckmark/cycle7/fixtures.py | 54 +++ fuckmark/cycle7/instrumentation.py | 286 +++++++++++++ fuckmark/cycle7/ledger.py | 110 +++++ fuckmark/cycle7/registry.py | 31 ++ fuckmark/cycle7/whitespace_collapse.py | 87 ++++ fuckmark/cycle7_stage_a_hf.py | 391 ++++++++++++++++++ fuckmark/transforms/cycle7_quote_policy.py | 130 ++++++ fuckmark/transforms/registry.py | 21 +- fuckmark/transforms/trace.py | 25 +- tests/test_cycle7_durable_rules.py | 139 +++++++ tests/test_cycle7_instrumentation.py | 83 ++++ tests/test_cycle7_ledger.py | 44 ++ tests/test_cycle7_quote_policy.py | 60 +++ tests/test_cycle7_sanitizer_boundary.py | 11 + tests/test_cycle7_whitespace_collapse.py | 48 +++ 32 files changed, 2412 insertions(+), 14 deletions(-) create mode 100644 docs/cycle7/README.md create mode 100644 docs/cycle7/literature.md create mode 100644 docs/cycle7/live-state-audit.md create mode 100644 docs/cycle7/seed-ledger.md create mode 100644 docs/cycle7/stage-a-decision.md create mode 100644 docs/cycle7/whitespace-collapse-v1.md create mode 100644 evidence/cycle7-stage-a-2026-08-25/README.md create mode 100644 evidence/cycle7-stage-a-2026-08-25/SHA256SUMS.txt create mode 100644 evidence/cycle7-stage-a-2026-08-25/decision-geometry-only.json create mode 100644 evidence/cycle7-stage-a-2026-08-25/decision.json create mode 100644 evidence/cycle7-stage-a-2026-08-25/detector-compare-family1-pre-copula.json create mode 100644 evidence/cycle7-stage-a-2026-08-25/detector-compare.json create mode 100644 evidence/cycle7-stage-a-2026-08-25/fixture-geometry.json create mode 100644 fuckmark/cycle7/__init__.py create mode 100644 fuckmark/cycle7/compare.py create mode 100644 fuckmark/cycle7/decision.py create mode 100644 fuckmark/cycle7/durable_rules.py create mode 100644 fuckmark/cycle7/fixtures.py create mode 100644 fuckmark/cycle7/instrumentation.py create mode 100644 fuckmark/cycle7/ledger.py create mode 100644 fuckmark/cycle7/registry.py create mode 100644 fuckmark/cycle7/whitespace_collapse.py create mode 100644 fuckmark/cycle7_stage_a_hf.py create mode 100644 fuckmark/transforms/cycle7_quote_policy.py create mode 100644 tests/test_cycle7_durable_rules.py create mode 100644 tests/test_cycle7_instrumentation.py create mode 100644 tests/test_cycle7_ledger.py create mode 100644 tests/test_cycle7_quote_policy.py create mode 100644 tests/test_cycle7_sanitizer_boundary.py create mode 100644 tests/test_cycle7_whitespace_collapse.py diff --git a/docs/cycle7/README.md b/docs/cycle7/README.md new file mode 100644 index 00000000..ff8f0a08 --- /dev/null +++ b/docs/cycle7/README.md @@ -0,0 +1,70 @@ +# Cycle 7 — durable transform development + +Cycle 6 is **frozen**. Formal outcome: `NONZERO_RESIDUAL` **7/192**, not 0/192. + +This directory starts Cycle 7 development. It does not retune Cycle 6 and does not authorize a new formal confirmation. + +## Live state (verified 2026-08-25) + +See `docs/cycle7/live-state-audit.md`. Independent GitHub inspection matched the Cycle 6 freeze: + +- `main` HEAD `ddccd74a9e0e710085b385fed98959f2730b9d60` +- PR #93 and PR #94 merged +- Recovery run `32886342498` success +- Aggregate hash `30577aafaffd0c50f0ddb384a4509eb0bb93e4374bf39704869ddbf5053186a4` + +## Spent-corpus rule + +Do not use as Cycle 7 development, tuning, or confirmation data: + +- `720000`, `730000` (earlier development) +- `760000`, `770000`, `780000` (formal confirmation; the 7 residuals are spent) + +The 7 formal residuals may be used only for high-level forensic classification. They must not be used as development examples. + +## Seed ledger + +See `docs/cycle7/seed-ledger.md` and `fuckmark/cycle7/ledger.py`. + +| Role | Seed base | Status | +| --- | --- | --- | +| Exploratory development (Stage A) | `810000` | in use | +| Validation development (Stage B) | `820000` | reserved, unused | +| Confirmation reserved | `830000`, `840000`, `850000` | must not be inspected | + +Confirmation seeds were chosen as unused 10k blocks before any Cycle 7 detector look. + +## Whitespace-collapse sanitizer + +See `docs/cycle7/whitespace-collapse-v1.md`. + +Cycle 7 evaluation arms: + +`raw`, `nfkc`, `cf_strip`, `nfkc_cf_strip`, `ws_collapse`, `ws_collapse_nfkc_cf_strip` + +The Cycle 6 frozen sanitizer module is unchanged. + +## Durable transform family (first attempt) + +Unambiguous contractions (existing catalog + reversible have-forms) and UK/US `towards`/`toward`, `amongst`/`among`. + +These change tokens without relying on repeated U+0020. They survive ASCII whitespace collapse by construction. + +Quote interiors may receive these durable edits under `quote-container-durable-v1` without changing quote delimiters. Cycle 6 `quote-container-surface-spacing-v1` still admits only spacing inside quotes. + +## Stage A result + +**`INSUFFICIENT_EVIDENCE`** as a Cycle 6 replacement. + +Cycle 6 spacing: 0/4 raw detections, 4/4 after whitespace collapse (fresh seed `810000`). + +Cycle 7 durable family: 4/4 raw and 4/4 after collapse. Collapse-survival works on contraction-rich fixtures; natural density on GPT-2 TinyDev text is too low. + +Details: `docs/cycle7/stage-a-decision.md`. + + +- Not a Cycle 7 formal confirmation +- Not a ZERO_RESIDUAL claim +- Not human-fidelity validation +- Not threshold or FPR manipulation +- Not neural rewriting diff --git a/docs/cycle7/literature.md b/docs/cycle7/literature.md new file mode 100644 index 00000000..5199a946 --- /dev/null +++ b/docs/cycle7/literature.md @@ -0,0 +1,24 @@ +# Cycle 7 literature notes + +Claim labels used below: VERIFIED, SOURCE-BOUND, HYPOTHESIS, UNKNOWN, REJECTED, EXTERNAL-VALIDATION-ONLY. + +## SynthID-Text + +SOURCE-BOUND. Google documents SynthID-Text as a logits-processor watermark using a keyed g-function and tournament sampling. Detectability uses n-gram context. Larger `ngram_len` is more detectable and more brittle to edits. Public docs state robustness to cropping, a few word changes, and mild paraphrasing, with large detector-confidence loss under thorough rewrite or translation. + +- https://ai.google.dev/responsible/docs/safeguards/synthid +- https://huggingface.co/blog/synthid-text + +This repository's Cycle 6 result is the local VERIFIED measurement for the frozen GPT-2 / mean-score protocol: 183/192 pristine detections reduced to 7/192 transformed detections, 0/192 transformed unwatermarked, `NONZERO_RESIDUAL`. That is not a claim about proprietary production detectors. + +## Edit-channel / synchronization + +SOURCE-BOUND / HYPOTHESIS. Context-hashed watermarks re-synchronize after a local edit once the sliding n-gram window has moved past the edit. Cycle 6 spacing inserts extra U+0020 tokens that change GPT-2 segmentation, then ordinary space-collapse removes those tokens and restores the original token sequence. That collapse failure is VERIFIED here as a mechanism class, not as a universal watermark theorem. + +EXTERNAL-VALIDATION-ONLY. Zhao et al. style analyses of SynthID robustness (paraphrase, back-translation, layer-inflation) are not reproduced in this repository and must not be cited as FuckMark results. + +## Deterministic non-neural edits + +HYPOTHESIS. Unambiguous contractions and closed orthographic variants change tokenizer boundaries without depending on repeated spaces. Whether they destroy enough SynthID observation windows on GPT-2 text is a Cycle 7 measurement, not a literature conclusion. + +REJECTED as a Cycle 7 direction: adding more U+0020 spaces to chase 7/192 → 0/192 on spent confirmation rows. diff --git a/docs/cycle7/live-state-audit.md b/docs/cycle7/live-state-audit.md new file mode 100644 index 00000000..f897b163 --- /dev/null +++ b/docs/cycle7/live-state-audit.md @@ -0,0 +1,57 @@ +# Cycle 7 live-state audit + +Audit time: 2026-08-25 (America/New_York). Live GitHub overrides any stale handoff. + +## Repository + +- Remote: `https://github.com/byte271/FuckMark` +- Verified `main` HEAD: `ddccd74a9e0e710085b385fed98959f2730b9d60` +- Commit subject: `Freeze Cycle 6 formal NONZERO_RESIDUAL confirmation evidence (#94)` +- Committer date: 2026-08-25 19:07:20 UTC + +Local `origin/main` matched this SHA after `git fetch`. + +## Pull requests + +| PR | State | Merged | Merge commit | +| --- | --- | --- | --- | +| #93 Recover Cycle 6 run 16 freeze artifacts | closed | yes, 2026-08-25 18:52:36 UTC | `1e86be770ae231046055647689a4836d221d8274` | +| #94 Freeze Cycle 6 formal NONZERO_RESIDUAL evidence | closed | yes, 2026-08-25 19:07:20 UTC | `ddccd74a9e0e710085b385fed98959f2730b9d60` | + +Open PRs at audit time: **none**. +Open issues at audit time: **none**. + +PR #93 had CodeRabbit review comments (3). PR #94 had no review comments. Both PRs are merged; those threads are not open Cycle 7 blockers. + +## GitHub Actions + +| Run | Workflow | Conclusion | +| --- | --- | --- | +| `32873260399` | Cycle6 Sealed Confirmation (original freeze) | **failure** (cross-check after freeze; science already frozen) | +| `32886342498` | Cycle6 Frozen Artifact Recovery | **success** | +| `32887767052` | CI on `main` @ `ddccd74` | **success** | +| `32887767017` | Release Engineering on `main` | **success** | +| `32887767000` | Release Readiness Baseline on `main` | **success** | + +Handoff vs live GitHub: **no contradiction**. + +## Cycle 6 frozen evidence (rehashed locally) + +Directory: `evidence/cycle6-sealed-confirmation-recovery-2026-08-25/` + +SHA256SUMS.txt verified for every listed file. + +| Binding | Verified value | +| --- | --- | +| Formal outcome | `NONZERO_RESIDUAL` | +| Pooled transformed WM | 7/192 on all four frozen sanitizer arms | +| Pooled transformed UW | 0/192 | +| Aggregate hash | `30577aafaffd0c50f0ddb384a4509eb0bb93e4374bf39704869ddbf5053186a4` | +| Contract hash | `8bff80151c1be33a9f4bedf0b00abab1fffd9b04c0572aef1381be58530e1cef` | +| Recovery cross-check hash | `8a65367d6aebecd26028150a58722c69cd47deb55df15e7dca24a56366306207` | +| Recovery manifest hash | `d0639498103b2322576c99108d00a3eed7a6d99f317491fab558e6b49055ff99` | +| Scientific commit | `bfd9a4d81f0561a17f5ac4daa3858e97ebd811f1` | +| Recovery orchestration commit | `1e86be770ae231046055647689a4836d221d8274` | +| Original freeze artifacts 760/770/780 | IDs and digests match the recovery manifest | + +Do not describe Cycle 6 as `ZERO_RESIDUAL`. diff --git a/docs/cycle7/seed-ledger.md b/docs/cycle7/seed-ledger.md new file mode 100644 index 00000000..af8b35a3 --- /dev/null +++ b/docs/cycle7/seed-ledger.md @@ -0,0 +1,26 @@ +# Cycle 7 seed ledger + +Ledger identity: `cycle7-seed-ledger-v1` (`fuckmark/cycle7/ledger.py`). + +Seeds were assigned **before** Cycle 7 detector scoring. + +## Spent (forbidden) + +| Seed | Why | +| --- | --- | +| 760000, 770000, 780000 | Cycle 6 formal confirmation. Spent. | +| 720000, 730000 | Cycle 6 development 0/16. Spent. | +| 401000, 402000, 500000, 61000 | Historic TinyDev / calibration / schedule bases | +| 1120000–1160000 | Frozen effectiveness-profile schedule bases | + +## Cycle 7 roles + +| Role | Seed | May score detectors? | +| --- | --- | --- | +| Exploratory development (Stage A) | 810000 | yes, development only | +| Validation development (Stage B) | 820000 | only after Stage A family freeze; not in this PR's detector run | +| Confirmation reserved | 830000, 840000, 850000 | **no** until a later frozen confirmation protocol | + +Do not promote 810000 or 820000 into confirmation after seeing scores. + +Stage A generation uses 810000 with pair stride 32, matching TinyDev pairing, one pair per domain, topic `reproducibility` chosen without a detector look. diff --git a/docs/cycle7/stage-a-decision.md b/docs/cycle7/stage-a-decision.md new file mode 100644 index 00000000..0fe6c3c7 --- /dev/null +++ b/docs/cycle7/stage-a-decision.md @@ -0,0 +1,52 @@ +# Cycle 7 Stage A decision + +**Decision: `INSUFFICIENT_EVIDENCE`** + +This is development-only. It is not a Cycle 7 formal confirmation. Seeds `830000` / `840000` / `850000` were not inspected. + +## What was measured + +Fresh exploratory seed `810000`, one pair per domain (8 texts), frozen GPT-2 revision, frozen Cycle 6 threshold `0.5570987654320988`, budget 14, detector-blind cover-greedy v4. + +Arms on the same texts: + +- Cycle 6 spacing-heavy B14 (`quote_safe_zrd_transform_registry`) +- Cycle 7 durable (contractions + bounded `it's` copulas + toward/among; no extra spaces) +- Combined durable + Cycle 6 spacing + +Sanitizer arms included `raw` and `ws_collapse`. + +## Detector counts (watermarked 4, unwatermarked 4) + +| Arm | Pristine WM | Raw WM | Raw UW | ws_collapse WM | ws_collapse UW | +| --- | --- | --- | --- | --- | --- | +| Cycle 6 spacing B14 | 4/4 | **0/4** | 0/4 | **4/4** | 0/4 | +| Cycle 7 durable | 4/4 | **4/4** | 0/4 | **4/4** | 0/4 | +| Combined | 4/4 | **0/4** | 0/4 | **4/4** | 0/4 | + +Family-1 snapshot (before bounded `it's` copulas) is preserved as `detector-compare-family1-pre-copula.json`. After adding `it's like` / `it's important` / `it's a|an|the`, durable still 4/4 detected. Two watermarked samples gained surviving non-space edits (`it's important`, `it's like`); that did not move detections under the frozen threshold. + +## VERIFIED (this Stage A) + +1. Cycle 6 spacing can drive raw detections from 4/4 to 0/4 on this tiny corpus. +2. The same Cycle 6 outputs are detected 4/4 after `whitespace-collapse-v1`. Collapse restores the attack. This is the Cycle 6 channel weakness, reproduced on **new** development data, not on spent 760000–780000 rows. +3. Unwatermarked transformed controls stayed 0/4 on every arm. +4. On contraction-rich fixtures, durable edits remain after collapse; inside quotes, Cycle 6 spacing collapses away and Cycle 7 durable edits remain. +5. Bare `it's not` ↔ `it is not` is not admitted: hard invariants treat `it's not` as a different negation atom than `is not`, matching the existing `he's not` / `he is not` split. + +## SOURCE-BOUND forensic note (Cycle 6 spent residuals) + +Not used for tuning. High-level class only: most formal residuals looked like budget-limited leftover leak windows under a spacing-heavy attack. + +## HYPOTHESIS going forward + +GPT-2 64-token TinyDev continuations rarely contain reversible contraction/orthography sites. A collapse-resistant family needs **higher natural density** than unambiguous copula/orthography, or a different tokenizer-boundary channel that is still non-neural and detector-blind. + +Do not chase 7/192 → 0/192 by adding more U+0020 spaces. + +## Next development direction + +1. Keep Cycle 6 geometry/scheduler/protected-span machinery. +2. Search additional **high-frequency, invariant-safe** durable rules (not ambiguous `'d` / bare `it's`). +3. Measure density on Stage B seed `820000` only after a family is frozen. +4. Do not open Cycle 7 formal confirmation until a collapse-resistant family replicates on disjoint development data. diff --git a/docs/cycle7/whitespace-collapse-v1.md b/docs/cycle7/whitespace-collapse-v1.md new file mode 100644 index 00000000..1aa24285 --- /dev/null +++ b/docs/cycle7/whitespace-collapse-v1.md @@ -0,0 +1,27 @@ +# whitespace-collapse-v1 + +Sanitizer identity: `ascii-horizontal-run-collapse-preserve-newlines-v1`. + +This is a **new** Cycle 7 sanitizer. It does not modify frozen `fuckmark/sanitizer_robustness.py`. + +## Exact behavior + +1. Map CR LF and CR to LF. +2. Preserve LF. Do not join lines. +3. On each line, replace every maximal run of U+0020 SPACE and U+0009 HT with one U+0020 SPACE. +4. Do not trim a leftover single leading or trailing space. +5. Do not collapse other Unicode whitespace, including NBSP (U+00A0). +6. Do not strip Cf characters (that is a separate arm). + +## Cycle 7 variant IDs + +| ID | Definition | +| --- | --- | +| `raw` | identity | +| `nfkc` | Unicode NFKC (imported from the frozen Cycle 6 helper) | +| `cf_strip` | drop Unicode category Cf | +| `nfkc_cf_strip` | Cf-strip after NFKC | +| `ws_collapse` | whitespace-collapse-v1 | +| `ws_collapse_nfkc_cf_strip` | collapse, then NFKC, then Cf-strip, then collapse again so NFKC-introduced ASCII spaces also collapse | + +Whitespace robustness may be claimed only if the `ws_collapse` and `ws_collapse_nfkc_cf_strip` arms pass. diff --git a/evidence/cycle7-stage-a-2026-08-25/README.md b/evidence/cycle7-stage-a-2026-08-25/README.md new file mode 100644 index 00000000..5419003a --- /dev/null +++ b/evidence/cycle7-stage-a-2026-08-25/README.md @@ -0,0 +1,25 @@ +# Cycle 7 Stage A evidence + +Development-only. Not a formal confirmation. Seed `810000` is the exploratory ledger entry. + +Files: + +- `fixture-geometry.json` — OffsetTokenizer geometry on hand fixtures, plus detector summaries when scored. +- `decision-geometry-only.json` — classification from fixtures before detector scoring (`PROMISING_DEVELOPMENT` for the collapse-survival *channel*). +- `detector-compare-family1-pre-copula.json` — first durable catalog on the generated corpus (almost no sites). +- `detector-compare.json` — same generated texts rescored after bounded `it's` copulas. +- `decision.json` — Stage A decision after detector attachment: `INSUFFICIENT_EVIDENCE`. + +Do not treat these rows as confirmation data. Do not reuse them to retune Cycle 6. + + +Development-only. Not a formal confirmation. Seed `810000` is the exploratory ledger entry. + +Files: + +- `fixture-geometry.json` — OffsetTokenizer (byte) geometry on hand fixtures. CI-reproducible. +- `decision-geometry-only.json` — classification before detector scoring. +- `detector-compare.json` — GPT-2 / frozen threshold scoring on a fresh 8-sample corpus (added if the local HF run completes). +- `decision.json` — final Stage A decision after detector attachment when available. + +Do not treat these rows as confirmation data. Do not reuse them to retune Cycle 6. diff --git a/evidence/cycle7-stage-a-2026-08-25/SHA256SUMS.txt b/evidence/cycle7-stage-a-2026-08-25/SHA256SUMS.txt new file mode 100644 index 00000000..99004ae6 --- /dev/null +++ b/evidence/cycle7-stage-a-2026-08-25/SHA256SUMS.txt @@ -0,0 +1,6 @@ +0fe1edfb1272911ba4a0052a67235edef58237483685acf21f122872e5a914b4 decision-geometry-only.json +f2812647a9f5fcb13738947d6b613cf32af0893bf8c2a621c0c14e7273c2c1bf decision.json +84082cde1c2bf42911f637913dd0eb1e64f94529f1a1727dc4ad0b94fad7d63d detector-compare-family1-pre-copula.json +4bf03dcdebdb39b52a62edb31dc53b811e53635bad1e4038a23e269f319ee760 detector-compare.json +fc9bdc64db0639410ff3cfbaa18b6009c6e265a2913e4d04dab8361108c4eeff fixture-geometry.json +811257e314caedf3c94dfcafc3cc7943f0662e79b66f07a46b7c8cfa41e2a878 README.md diff --git a/evidence/cycle7-stage-a-2026-08-25/decision-geometry-only.json b/evidence/cycle7-stage-a-2026-08-25/decision-geometry-only.json new file mode 100644 index 00000000..89486eeb --- /dev/null +++ b/evidence/cycle7-stage-a-2026-08-25/decision-geometry-only.json @@ -0,0 +1 @@ +{"algorithm_version":"cycle7-stage-a-decision-v1","decision":"PROMISING_DEVELOPMENT","detector_signal":null,"notes":"This is a development classification, not a Cycle 7 formal confirmation. Confirmation seeds 830000/840000/850000 were not inspected.","reasons":["durable edits remain after ASCII whitespace collapse on contraction-rich text","inside quotes, Cycle 6 spacing collapses away while Cycle 7 durable edits remain","durable candidate density is zero on contraction-sparse text","combined durable+spacing still enumerates a non-empty candidate set","geometry-only Stage A; detector scoring was not attached"]} diff --git a/evidence/cycle7-stage-a-2026-08-25/decision.json b/evidence/cycle7-stage-a-2026-08-25/decision.json new file mode 100644 index 00000000..49514337 --- /dev/null +++ b/evidence/cycle7-stage-a-2026-08-25/decision.json @@ -0,0 +1 @@ +{"algorithm_version":"cycle7-stage-a-decision-v1","decision":"INSUFFICIENT_EVIDENCE","detector_signal":{"durable_raw_detected":4,"durable_ws_collapse_detected":4,"spacing_raw_detected":0,"spacing_ws_collapse_detected":4,"watermarked_row_count":4},"notes":"This is a development classification, not a Cycle 7 formal confirmation. Confirmation seeds 830000/840000/850000 were not inspected.","reasons":["durable edits remain after ASCII whitespace collapse on contraction-rich text","inside quotes, Cycle 6 spacing collapses away while Cycle 7 durable edits remain","durable candidate density is zero on contraction-sparse text","combined durable+spacing still enumerates a non-empty candidate set","Cycle 6 spacing reduced raw detections and those detections returned after whitespace collapse","durable family did not reduce raw detections on the Stage A corpus","durable collapse-survival is real on fixtures, but density is too low to replace Cycle 6 spacing"]} diff --git a/evidence/cycle7-stage-a-2026-08-25/detector-compare-family1-pre-copula.json b/evidence/cycle7-stage-a-2026-08-25/detector-compare-family1-pre-copula.json new file mode 100644 index 00000000..5919405f --- /dev/null +++ b/evidence/cycle7-stage-a-2026-08-25/detector-compare-family1-pre-copula.json @@ -0,0 +1 @@ +{"algorithm_version":"cycle7-stage-a-detector-compare-v1","artifact_hash":"5e838c3c62fc5b7de657631f875745474a77ce69b2dbac414a67264561eea424","budget":14,"detector_access_used_for_selection":false,"geometry_rows":[{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"db8d7d1137950ec78c16db2aeba5ce6cb1fe2ef9d1f539bb04496aa0306d4b02","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"34484f77ef9a6549354b3b2baac224dce1cf6d0bb147445e0dadaaed77032d63","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":39,"measurement_hash":"3e7686ab59a4667f19eb7add5915410bad8c717cfea5dd7a5b2c636fbc8c8ea9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":39,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":39}},"domain":"general_explanatory","label":"unwatermarked","sample_id":"cycle7-stage-a-general_explanatory-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"54cd460f7640946d165a3d13cf2cf566e09f24c424f5eb16e30aa1768e57e2ed","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"7539cea90f14a89279376319325bdb3c41f08167920510258de2b4127175dca4","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":60,"measurement_hash":"383ef78436d4edb8f4d761596ebbe838a527973f6bfd5735fe319d5145ce8adf","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":60}},"domain":"general_explanatory","label":"watermarked","sample_id":"cycle7-stage-a-general_explanatory-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":4,"measurement_hash":"ea94965fd21cb116e8b37ecf1bf099857b9244d663d9e28feacc6f21e9632ad0","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-comma","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":7},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":4,"measurement_hash":"ad231010c24e615a98ddb91a2baad454c1ef4907de65fa662fbbb85cf11c2514","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-comma","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":7},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":40,"measurement_hash":"e1a07d3e84c390d10423916b1c0d7cdff68e9f38d2b413c2afd8e51b163cba5f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":40,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":40}},"domain":"technical_explanation","label":"unwatermarked","sample_id":"cycle7-stage-a-technical_explanation-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"a95ed6ff89d38bb3f97a1fe57ad8f8cd6b5f6af5fd667067773e41f6782bf373","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"93cb2a3afe25062bedd167067e8cf70c4c6e57e372fd7e6e30235670d5275db7","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":54,"measurement_hash":"7a5fd65de3dcdeb0322f31f88d1b24c796f6cc421d36494d9049403fe493a24a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":54,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":54}},"domain":"technical_explanation","label":"watermarked","sample_id":"cycle7-stage-a-technical_explanation-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":57,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":57,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":8,"measurement_hash":"93b4df80d55e224ac27b52995580fe92743ff437183dbcab8c52d613d64cf237","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":9},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":53,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":46,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":7,"measurement_hash":"aeaf328f84e96fddc16b96dd6850fe96a4ec58760c6bfb0fb9848992c98920ef","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","expand-you-are","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","expand-that-is-not-x","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":12,"tuple_leak_window_count":8},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":4,"candidate_exhausted":true,"changed_token_positions":5,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion"],"intact_window_count":40,"measurement_hash":"594f2148dd04a002541ac9bdd0a63a6aaeb1cb3fbc0e2a20dfbf6343d306146a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":57,"selected_count":3,"selected_rule_ids":["expand-you-are","expand-do-not","expand-that-is-not-x"],"token_delta":0,"tuple_leak_window_count":40}},"domain":"conversational_prose","label":"unwatermarked","sample_id":"cycle7-stage-a-conversational_prose-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"fa9c76c6ddb778141a4962a77e8002c1413ca7289d88c89177fe9d861683a4b5","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"3868bda9994b005506d125b7a70b2110f37b3b27cdb51c088ee19cd636b20c71","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"a815a53516b54ea1011be7ad4d6e8fc09b998a1bd7770f4c7171fad6431e4c4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59}},"domain":"conversational_prose","label":"watermarked","sample_id":"cycle7-stage-a-conversational_prose-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"c89d29f8c4ff8d8b907f91b255343f24aebb113f6f990e9dba00911566095e4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"9483f967307734bda93606f643cfad5f2ad1632797995b994ce16324a0dbfffc","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"d8e63ad3f8cddab8110faa1c91ecd15fc2161d622d41d3a655252553eb70c92a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59}},"domain":"structured_instructional","label":"unwatermarked","sample_id":"cycle7-stage-a-structured_instructional-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":44,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":13,"measurement_hash":"4fa592661bb841d8a40f7403a54f3c091faf666d726a680b733527f54d98a981","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":29,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma"],"token_delta":14,"tuple_leak_window_count":13},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":46,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":49,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":49,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":11,"measurement_hash":"205d8b8727464d4770d419ee86b3f5d728b3d2c353363dfd8bb139189a054743","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":29,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","expand-cannot","surface-space-after-any-word","cycle7-expand-i-have","surface-space-after-comma"],"token_delta":11,"tuple_leak_window_count":11},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":2,"candidate_exhausted":true,"changed_token_positions":17,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":49,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":49,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":49,"measurement_hash":"c1a0a69039873782e1c8520eb8c0d78fd44fba03ba57901dbe5bf2ed451f132c","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":2,"selected_rule_ids":["expand-cannot","cycle7-expand-i-have"],"token_delta":-1,"tuple_leak_window_count":49}},"domain":"structured_instructional","label":"watermarked","sample_id":"cycle7-stage-a-structured_instructional-watermarked"}],"model":"openai-community/gpt2","model_revision":"607a30d783dfa663caf39e06633721c8d4cfcd7e","samples":[{"domain":"general_explanatory","label":"unwatermarked","sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","seed":810000,"text":"\n\nRead the documentation.\n\nWrite a proof of concept and a proof of proof of hypothesis.\n\nWrite a proof of concept and proof of hypothesis for a variety of reasons.\n\nWrite a proof of concept and proof of hypothesis for a variety of reasons. Create a proof of hypothesis and proof of hypothesis","text_sha256":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},{"domain":"general_explanatory","label":"watermarked","sample_id":"cycle7-stage-a-general_explanatory-watermarked","seed":810000,"text":" Use a scientific term for the same thing.\n\nHow do I keep track of my data?\n\nIt's not uncommon for someone to want to know if a new test set is working or not. I use a lot of tools and data sources to keep track of this information. It's important to keep your","text_sha256":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},{"domain":"technical_explanation","label":"unwatermarked","sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","seed":810033,"text":" Then, make a test suite. If the software fails, give the user a warning, but otherwise give the user an option to try again. If the software fails, give the user a warning, but otherwise give the user an option to try again. In a test suite, it's an important consideration that if the","text_sha256":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},{"domain":"technical_explanation","label":"watermarked","sample_id":"cycle7-stage-a-technical_explanation-watermarked","seed":810033,"text":" Check the validity of the hypothesis.\n\nTest for reproducibility\n\nFigure 3.5 (b) The following table summarizes the steps involved in reproducibility of the hypothesis.\n\n(a) Proof-of-concept (ROS)\n\nTo test reproducibility of the hypothesis, a","text_sha256":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},{"domain":"conversational_prose","label":"unwatermarked","sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","seed":810064,"text":"\n\nIn fact, when you're trying to answer a question, you may need to clarify the question to avoid confusion. Don't make a mistake that isn't relevant to the question.\n\nIn a nutshell:\n\nYou need to answer a question properly.\n\nYou need to give your answer in a","text_sha256":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"},{"domain":"conversational_prose","label":"watermarked","sample_id":"cycle7-stage-a-conversational_prose-watermarked","seed":810064,"text":"\n\nTry:\n\n1. Introduce a point of view, or question, in your own context.\n\n2. Discuss the importance of \"evidence\".\n\n3. Give a strong critique of your own research.\n\n4. Use the same words or phrases in your own work.\n\n","text_sha256":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},{"domain":"structured_instructional","label":"unwatermarked","sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","seed":810096,"text":"\n\nAn experimenter's description of the method is in the section below.\n\nThe test can be run on a computer or an operating system, using the \"C\" or \"F\" buttons. You can use the \"R\" or \"D\" buttons to control the program.\n\nThe program is","text_sha256":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},{"domain":"structured_instructional","label":"watermarked","sample_id":"cycle7-stage-a-structured_instructional-watermarked","seed":810096,"text":" Use a series of simple examples to illustrate the results.\n\n\"It's like having an elephant.\"\n\nA lot of people have asked me about this: \"How do you teach a learner to write a sentence that can't be changed at all?\"\n\nWhat I've learned, however, is that","text_sha256":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"}],"sanitizer_ids":["raw","nfkc","cf_strip","nfkc_cf_strip","ws_collapse","ws_collapse_nfkc_cf_strip"],"scored_rows":[{"arm_id":"cycle6_spacing_b14","domain":"general_explanatory","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"db8d7d1137950ec78c16db2aeba5ce6cb1fe2ef9d1f539bb04496aa0306d4b02","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc_cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"raw":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"ws_collapse":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"}}},{"arm_id":"cycle7_durable","domain":"general_explanatory","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":39,"measurement_hash":"3e7686ab59a4667f19eb7add5915410bad8c717cfea5dd7a5b2c636fbc8c8ea9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":39,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":39},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"nfkc":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"raw":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"}}},{"arm_id":"cycle7_combined","domain":"general_explanatory","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"34484f77ef9a6549354b3b2baac224dce1cf6d0bb147445e0dadaaed77032d63","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc_cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"raw":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"ws_collapse":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"}}},{"arm_id":"cycle6_spacing_b14","domain":"general_explanatory","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"54cd460f7640946d165a3d13cf2cf566e09f24c424f5eb16e30aa1768e57e2ed","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc_cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"raw":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"ws_collapse":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"}}},{"arm_id":"cycle7_durable","domain":"general_explanatory","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":60,"measurement_hash":"383ef78436d4edb8f4d761596ebbe838a527973f6bfd5735fe319d5145ce8adf","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":60},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"nfkc":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"nfkc_cf_strip":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"raw":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"ws_collapse":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"}}},{"arm_id":"cycle7_combined","domain":"general_explanatory","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"7539cea90f14a89279376319325bdb3c41f08167920510258de2b4127175dca4","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc_cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"raw":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"ws_collapse":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"}}},{"arm_id":"cycle6_spacing_b14","domain":"technical_explanation","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":4,"measurement_hash":"ea94965fd21cb116e8b37ecf1bf099857b9244d663d9e28feacc6f21e9632ad0","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-comma","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":7},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"nfkc":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"nfkc_cf_strip":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"raw":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"ws_collapse":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"}}},{"arm_id":"cycle7_durable","domain":"technical_explanation","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":40,"measurement_hash":"e1a07d3e84c390d10423916b1c0d7cdff68e9f38d2b413c2afd8e51b163cba5f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":40,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":40},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},"nfkc":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},"nfkc_cf_strip":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},"raw":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},"ws_collapse":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"}}},{"arm_id":"cycle7_combined","domain":"technical_explanation","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":4,"measurement_hash":"ad231010c24e615a98ddb91a2baad454c1ef4907de65fa662fbbb85cf11c2514","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-comma","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":7},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"nfkc":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"nfkc_cf_strip":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"raw":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"ws_collapse":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"}}},{"arm_id":"cycle6_spacing_b14","domain":"technical_explanation","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"a95ed6ff89d38bb3f97a1fe57ad8f8cd6b5f6af5fd667067773e41f6782bf373","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc_cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"raw":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"ws_collapse":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"}}},{"arm_id":"cycle7_durable","domain":"technical_explanation","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":54,"measurement_hash":"7a5fd65de3dcdeb0322f31f88d1b24c796f6cc421d36494d9049403fe493a24a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":54,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":54},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"nfkc":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"raw":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"}}},{"arm_id":"cycle7_combined","domain":"technical_explanation","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"93cb2a3afe25062bedd167067e8cf70c4c6e57e372fd7e6e30235670d5275db7","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc_cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"raw":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"ws_collapse":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"}}},{"arm_id":"cycle6_spacing_b14","domain":"conversational_prose","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":57,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":57,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":8,"measurement_hash":"93b4df80d55e224ac27b52995580fe92743ff437183dbcab8c52d613d64cf237","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":9},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"nfkc":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"nfkc_cf_strip":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"raw":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"ws_collapse":{"detected":false,"score":0.5045631756158072,"text_hash":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5045631756158072,"text_hash":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"}}},{"arm_id":"cycle7_durable","domain":"conversational_prose","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":4,"candidate_exhausted":true,"changed_token_positions":5,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion"],"intact_window_count":40,"measurement_hash":"594f2148dd04a002541ac9bdd0a63a6aaeb1cb3fbc0e2a20dfbf6343d306146a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":57,"selected_count":3,"selected_rule_ids":["expand-you-are","expand-do-not","expand-that-is-not-x"],"token_delta":0,"tuple_leak_window_count":40},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"nfkc":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"nfkc_cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"raw":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"ws_collapse":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"}}},{"arm_id":"cycle7_combined","domain":"conversational_prose","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":53,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":46,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":7,"measurement_hash":"aeaf328f84e96fddc16b96dd6850fe96a4ec58760c6bfb0fb9848992c98920ef","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","expand-you-are","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","expand-that-is-not-x","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":12,"tuple_leak_window_count":8},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"nfkc":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"nfkc_cf_strip":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"raw":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"ws_collapse":{"detected":false,"score":0.4962785752259436,"text_hash":"9d238528d87390bd78adb434049836f4b4c913567ff043a539c90e5b4128c06d"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.4962785752259436,"text_hash":"9d238528d87390bd78adb434049836f4b4c913567ff043a539c90e5b4128c06d"}}},{"arm_id":"cycle6_spacing_b14","domain":"conversational_prose","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"fa9c76c6ddb778141a4962a77e8002c1413ca7289d88c89177fe9d861683a4b5","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc_cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"raw":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"ws_collapse":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"}}},{"arm_id":"cycle7_durable","domain":"conversational_prose","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"a815a53516b54ea1011be7ad4d6e8fc09b998a1bd7770f4c7171fad6431e4c4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"nfkc":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"raw":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"}}},{"arm_id":"cycle7_combined","domain":"conversational_prose","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"3868bda9994b005506d125b7a70b2110f37b3b27cdb51c088ee19cd636b20c71","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc_cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"raw":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"ws_collapse":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"}}},{"arm_id":"cycle6_spacing_b14","domain":"structured_instructional","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"c89d29f8c4ff8d8b907f91b255343f24aebb113f6f990e9dba00911566095e4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc_cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"raw":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle7_durable","domain":"structured_instructional","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"d8e63ad3f8cddab8110faa1c91ecd15fc2161d622d41d3a655252553eb70c92a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"nfkc":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"raw":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle7_combined","domain":"structured_instructional","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"9483f967307734bda93606f643cfad5f2ad1632797995b994ce16324a0dbfffc","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc_cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"raw":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle6_spacing_b14","domain":"structured_instructional","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":44,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":13,"measurement_hash":"4fa592661bb841d8a40f7403a54f3c091faf666d726a680b733527f54d98a981","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":29,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma"],"token_delta":14,"tuple_leak_window_count":13},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"nfkc":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"nfkc_cf_strip":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"raw":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"ws_collapse":{"detected":true,"score":0.653030303030303,"text_hash":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.653030303030303,"text_hash":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"}}},{"arm_id":"cycle7_durable","domain":"structured_instructional","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":2,"candidate_exhausted":true,"changed_token_positions":17,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":49,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":49,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":49,"measurement_hash":"c1a0a69039873782e1c8520eb8c0d78fd44fba03ba57901dbe5bf2ed451f132c","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":2,"selected_rule_ids":["expand-cannot","cycle7-expand-i-have"],"token_delta":-1,"tuple_leak_window_count":49},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6196712891628147,"text_hash":"ef8a6e8ad10982d60945c60b66f4eff587325eccd044738417c7c6e7f0c812f9"},"nfkc":{"detected":true,"score":0.6196712891628147,"text_hash":"ef8a6e8ad10982d60945c60b66f4eff587325eccd044738417c7c6e7f0c812f9"},"nfkc_cf_strip":{"detected":true,"score":0.6196712891628147,"text_hash":"ef8a6e8ad10982d60945c60b66f4eff587325eccd044738417c7c6e7f0c812f9"},"raw":{"detected":true,"score":0.6196712891628147,"text_hash":"ef8a6e8ad10982d60945c60b66f4eff587325eccd044738417c7c6e7f0c812f9"},"ws_collapse":{"detected":true,"score":0.6196712891628147,"text_hash":"ef8a6e8ad10982d60945c60b66f4eff587325eccd044738417c7c6e7f0c812f9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6196712891628147,"text_hash":"ef8a6e8ad10982d60945c60b66f4eff587325eccd044738417c7c6e7f0c812f9"}}},{"arm_id":"cycle7_combined","domain":"structured_instructional","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":46,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":49,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":49,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":11,"measurement_hash":"205d8b8727464d4770d419ee86b3f5d728b3d2c353363dfd8bb139189a054743","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":29,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","expand-cannot","surface-space-after-any-word","cycle7-expand-i-have","surface-space-after-comma"],"token_delta":11,"tuple_leak_window_count":11},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5045881348698249,"text_hash":"eab01af1bb8402c239487c1ce6fa3405fad0bb0c48735d16729cb28ab954a571"},"nfkc":{"detected":false,"score":0.5045881348698249,"text_hash":"eab01af1bb8402c239487c1ce6fa3405fad0bb0c48735d16729cb28ab954a571"},"nfkc_cf_strip":{"detected":false,"score":0.5045881348698249,"text_hash":"eab01af1bb8402c239487c1ce6fa3405fad0bb0c48735d16729cb28ab954a571"},"raw":{"detected":false,"score":0.5045881348698249,"text_hash":"eab01af1bb8402c239487c1ce6fa3405fad0bb0c48735d16729cb28ab954a571"},"ws_collapse":{"detected":true,"score":0.6196712891628147,"text_hash":"ef8a6e8ad10982d60945c60b66f4eff587325eccd044738417c7c6e7f0c812f9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6196712891628147,"text_hash":"ef8a6e8ad10982d60945c60b66f4eff587325eccd044738417c7c6e7f0c812f9"}}}],"secret_access_used_for_selection":false,"seed_base":810000,"summaries":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5254488114855487,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6426760741438237},"cycle7_combined":{"arm_id":"cycle7_combined","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5235160608732206,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6343363206769516},"cycle7_durable":{"arm_id":"cycle7_durable","median_raw_watermarked_score":0.6367424242424242,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":4,"raw_watermarked_mean_score":0.6343363206769516,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.6634137990070192,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6343363206769516}},"threshold":0.5570987654320988} diff --git a/evidence/cycle7-stage-a-2026-08-25/detector-compare.json b/evidence/cycle7-stage-a-2026-08-25/detector-compare.json new file mode 100644 index 00000000..09e82894 --- /dev/null +++ b/evidence/cycle7-stage-a-2026-08-25/detector-compare.json @@ -0,0 +1 @@ +{"algorithm_version":"cycle7-stage-a-detector-compare-v1","artifact_hash":"4eaf4f34a855e7e46bd52ad2fc596bb1992a6639c483dc9958abe12e04808558","budget":14,"detector_access_used_for_selection":false,"durable_catalog_version":"cycle7-durable-rule-catalog-v1","geometry_rows":[{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"db8d7d1137950ec78c16db2aeba5ce6cb1fe2ef9d1f539bb04496aa0306d4b02","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"34484f77ef9a6549354b3b2baac224dce1cf6d0bb147445e0dadaaed77032d63","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":39,"measurement_hash":"3e7686ab59a4667f19eb7add5915410bad8c717cfea5dd7a5b2c636fbc8c8ea9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":39,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":39}},"domain":"general_explanatory","label":"unwatermarked","sample_id":"cycle7-stage-a-general_explanatory-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"54cd460f7640946d165a3d13cf2cf566e09f24c424f5eb16e30aa1768e57e2ed","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":59,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"6e8e88992a401f525e06faf4b74a5b68d12fc1bd33b97a40f2ae20b14137c1bd","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":55,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":55,"failure_classes":["candidate_exhaustion"],"intact_window_count":55,"measurement_hash":"9855d01e136f0355a45c2a1e5a274d7812b90eeb722e30e735cbc7e1412abb19","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-important"],"token_delta":0,"tuple_leak_window_count":55}},"domain":"general_explanatory","label":"watermarked","sample_id":"cycle7-stage-a-general_explanatory-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":4,"measurement_hash":"ea94965fd21cb116e8b37ecf1bf099857b9244d663d9e28feacc6f21e9632ad0","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-comma","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":7},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":59,"candidate_exhausted":false,"changed_token_positions":69,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":3,"measurement_hash":"63d019618be128cffadf6f2bfc7561ce8bb8b210ff783e2a0dd50e5be6591745","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","cycle7-expand-its-an","surface-space-after-any-word"],"token_delta":13,"tuple_leak_window_count":9},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["candidate_exhaustion"],"intact_window_count":35,"measurement_hash":"2a3fe2c19447deb585f7ccf6db0b11d339809cf3cda079734bc9229de5eeb98d","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":40,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-an"],"token_delta":0,"tuple_leak_window_count":35}},"domain":"technical_explanation","label":"unwatermarked","sample_id":"cycle7-stage-a-technical_explanation-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"a95ed6ff89d38bb3f97a1fe57ad8f8cd6b5f6af5fd667067773e41f6782bf373","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"93cb2a3afe25062bedd167067e8cf70c4c6e57e372fd7e6e30235670d5275db7","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":54,"measurement_hash":"7a5fd65de3dcdeb0322f31f88d1b24c796f6cc421d36494d9049403fe493a24a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":54,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":54}},"domain":"technical_explanation","label":"watermarked","sample_id":"cycle7-stage-a-technical_explanation-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":57,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":57,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":8,"measurement_hash":"93b4df80d55e224ac27b52995580fe92743ff437183dbcab8c52d613d64cf237","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":9},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":53,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":46,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":7,"measurement_hash":"aeaf328f84e96fddc16b96dd6850fe96a4ec58760c6bfb0fb9848992c98920ef","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","expand-you-are","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","expand-that-is-not-x","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":12,"tuple_leak_window_count":8},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":4,"candidate_exhausted":true,"changed_token_positions":5,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion"],"intact_window_count":40,"measurement_hash":"594f2148dd04a002541ac9bdd0a63a6aaeb1cb3fbc0e2a20dfbf6343d306146a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":57,"selected_count":3,"selected_rule_ids":["expand-you-are","expand-do-not","expand-that-is-not-x"],"token_delta":0,"tuple_leak_window_count":40}},"domain":"conversational_prose","label":"unwatermarked","sample_id":"cycle7-stage-a-conversational_prose-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"fa9c76c6ddb778141a4962a77e8002c1413ca7289d88c89177fe9d861683a4b5","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"3868bda9994b005506d125b7a70b2110f37b3b27cdb51c088ee19cd636b20c71","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"a815a53516b54ea1011be7ad4d6e8fc09b998a1bd7770f4c7171fad6431e4c4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59}},"domain":"conversational_prose","label":"watermarked","sample_id":"cycle7-stage-a-conversational_prose-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"c89d29f8c4ff8d8b907f91b255343f24aebb113f6f990e9dba00911566095e4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"9483f967307734bda93606f643cfad5f2ad1632797995b994ce16324a0dbfffc","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"d8e63ad3f8cddab8110faa1c91ecd15fc2161d622d41d3a655252553eb70c92a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59}},"domain":"structured_instructional","label":"unwatermarked","sample_id":"cycle7-stage-a-structured_instructional-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":44,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":13,"measurement_hash":"4fa592661bb841d8a40f7403a54f3c091faf666d726a680b733527f54d98a981","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":29,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma"],"token_delta":14,"tuple_leak_window_count":13},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":47,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":9,"measurement_hash":"26ce88d91c96c2cd4c2f98a4418a6b9d82afe975d7d9e9f897d1734cfb75c4af","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","cycle7-expand-its-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","expand-cannot","surface-space-after-any-word","cycle7-expand-i-have","surface-space-after-comma"],"token_delta":10,"tuple_leak_window_count":9},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":3,"candidate_exhausted":true,"changed_token_positions":18,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":44,"measurement_hash":"79aae0d2f1050d53e025cf30201df0f5f5cc6604cfd556a2384002c1df777cb9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":3,"selected_rule_ids":["cycle7-expand-its-like","expand-cannot","cycle7-expand-i-have"],"token_delta":-1,"tuple_leak_window_count":44}},"domain":"structured_instructional","label":"watermarked","sample_id":"cycle7-stage-a-structured_instructional-watermarked"}],"model":"openai-community/gpt2","model_revision":"607a30d783dfa663caf39e06633721c8d4cfcd7e","samples":[{"domain":"general_explanatory","label":"unwatermarked","sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","seed":810000,"text":"\n\nRead the documentation.\n\nWrite a proof of concept and a proof of proof of hypothesis.\n\nWrite a proof of concept and proof of hypothesis for a variety of reasons.\n\nWrite a proof of concept and proof of hypothesis for a variety of reasons. Create a proof of hypothesis and proof of hypothesis","text_sha256":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},{"domain":"general_explanatory","label":"watermarked","sample_id":"cycle7-stage-a-general_explanatory-watermarked","seed":810000,"text":" Use a scientific term for the same thing.\n\nHow do I keep track of my data?\n\nIt's not uncommon for someone to want to know if a new test set is working or not. I use a lot of tools and data sources to keep track of this information. It's important to keep your","text_sha256":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},{"domain":"technical_explanation","label":"unwatermarked","sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","seed":810033,"text":" Then, make a test suite. If the software fails, give the user a warning, but otherwise give the user an option to try again. If the software fails, give the user a warning, but otherwise give the user an option to try again. In a test suite, it's an important consideration that if the","text_sha256":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},{"domain":"technical_explanation","label":"watermarked","sample_id":"cycle7-stage-a-technical_explanation-watermarked","seed":810033,"text":" Check the validity of the hypothesis.\n\nTest for reproducibility\n\nFigure 3.5 (b) The following table summarizes the steps involved in reproducibility of the hypothesis.\n\n(a) Proof-of-concept (ROS)\n\nTo test reproducibility of the hypothesis, a","text_sha256":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},{"domain":"conversational_prose","label":"unwatermarked","sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","seed":810064,"text":"\n\nIn fact, when you're trying to answer a question, you may need to clarify the question to avoid confusion. Don't make a mistake that isn't relevant to the question.\n\nIn a nutshell:\n\nYou need to answer a question properly.\n\nYou need to give your answer in a","text_sha256":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"},{"domain":"conversational_prose","label":"watermarked","sample_id":"cycle7-stage-a-conversational_prose-watermarked","seed":810064,"text":"\n\nTry:\n\n1. Introduce a point of view, or question, in your own context.\n\n2. Discuss the importance of \"evidence\".\n\n3. Give a strong critique of your own research.\n\n4. Use the same words or phrases in your own work.\n\n","text_sha256":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},{"domain":"structured_instructional","label":"unwatermarked","sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","seed":810096,"text":"\n\nAn experimenter's description of the method is in the section below.\n\nThe test can be run on a computer or an operating system, using the \"C\" or \"F\" buttons. You can use the \"R\" or \"D\" buttons to control the program.\n\nThe program is","text_sha256":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},{"domain":"structured_instructional","label":"watermarked","sample_id":"cycle7-stage-a-structured_instructional-watermarked","seed":810096,"text":" Use a series of simple examples to illustrate the results.\n\n\"It's like having an elephant.\"\n\nA lot of people have asked me about this: \"How do you teach a learner to write a sentence that can't be changed at all?\"\n\nWhat I've learned, however, is that","text_sha256":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"}],"sanitizer_ids":["raw","nfkc","cf_strip","nfkc_cf_strip","ws_collapse","ws_collapse_nfkc_cf_strip"],"scored_rows":[{"arm_id":"cycle6_spacing_b14","domain":"general_explanatory","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"db8d7d1137950ec78c16db2aeba5ce6cb1fe2ef9d1f539bb04496aa0306d4b02","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc_cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"raw":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"ws_collapse":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"}}},{"arm_id":"cycle7_durable","domain":"general_explanatory","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":39,"measurement_hash":"3e7686ab59a4667f19eb7add5915410bad8c717cfea5dd7a5b2c636fbc8c8ea9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":39,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":39},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"nfkc":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"raw":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"}}},{"arm_id":"cycle7_combined","domain":"general_explanatory","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"34484f77ef9a6549354b3b2baac224dce1cf6d0bb147445e0dadaaed77032d63","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc_cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"raw":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"ws_collapse":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"}}},{"arm_id":"cycle6_spacing_b14","domain":"general_explanatory","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"54cd460f7640946d165a3d13cf2cf566e09f24c424f5eb16e30aa1768e57e2ed","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc_cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"raw":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"ws_collapse":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"}}},{"arm_id":"cycle7_durable","domain":"general_explanatory","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":55,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":55,"failure_classes":["candidate_exhaustion"],"intact_window_count":55,"measurement_hash":"9855d01e136f0355a45c2a1e5a274d7812b90eeb722e30e735cbc7e1412abb19","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-important"],"token_delta":0,"tuple_leak_window_count":55},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"nfkc":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"nfkc_cf_strip":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"raw":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"ws_collapse":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"}}},{"arm_id":"cycle7_combined","domain":"general_explanatory","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":59,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"6e8e88992a401f525e06faf4b74a5b68d12fc1bd33b97a40f2ae20b14137c1bd","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc_cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"raw":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"ws_collapse":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"}}},{"arm_id":"cycle6_spacing_b14","domain":"technical_explanation","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":4,"measurement_hash":"ea94965fd21cb116e8b37ecf1bf099857b9244d663d9e28feacc6f21e9632ad0","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-comma","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":7},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"nfkc":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"nfkc_cf_strip":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"raw":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"ws_collapse":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"}}},{"arm_id":"cycle7_durable","domain":"technical_explanation","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["candidate_exhaustion"],"intact_window_count":35,"measurement_hash":"2a3fe2c19447deb585f7ccf6db0b11d339809cf3cda079734bc9229de5eeb98d","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":40,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-an"],"token_delta":0,"tuple_leak_window_count":35},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"nfkc":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"nfkc_cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"raw":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"ws_collapse":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"}}},{"arm_id":"cycle7_combined","domain":"technical_explanation","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":59,"candidate_exhausted":false,"changed_token_positions":69,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":3,"measurement_hash":"63d019618be128cffadf6f2bfc7561ce8bb8b210ff783e2a0dd50e5be6591745","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","cycle7-expand-its-an","surface-space-after-any-word"],"token_delta":13,"tuple_leak_window_count":9},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"nfkc":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"nfkc_cf_strip":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"raw":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"ws_collapse":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"}}},{"arm_id":"cycle6_spacing_b14","domain":"technical_explanation","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"a95ed6ff89d38bb3f97a1fe57ad8f8cd6b5f6af5fd667067773e41f6782bf373","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc_cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"raw":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"ws_collapse":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"}}},{"arm_id":"cycle7_durable","domain":"technical_explanation","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":54,"measurement_hash":"7a5fd65de3dcdeb0322f31f88d1b24c796f6cc421d36494d9049403fe493a24a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":54,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":54},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"nfkc":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"raw":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"}}},{"arm_id":"cycle7_combined","domain":"technical_explanation","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"93cb2a3afe25062bedd167067e8cf70c4c6e57e372fd7e6e30235670d5275db7","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc_cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"raw":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"ws_collapse":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"}}},{"arm_id":"cycle6_spacing_b14","domain":"conversational_prose","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":57,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":57,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":8,"measurement_hash":"93b4df80d55e224ac27b52995580fe92743ff437183dbcab8c52d613d64cf237","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":9},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"nfkc":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"nfkc_cf_strip":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"raw":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"ws_collapse":{"detected":false,"score":0.5045631756158072,"text_hash":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5045631756158072,"text_hash":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"}}},{"arm_id":"cycle7_durable","domain":"conversational_prose","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":4,"candidate_exhausted":true,"changed_token_positions":5,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion"],"intact_window_count":40,"measurement_hash":"594f2148dd04a002541ac9bdd0a63a6aaeb1cb3fbc0e2a20dfbf6343d306146a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":57,"selected_count":3,"selected_rule_ids":["expand-you-are","expand-do-not","expand-that-is-not-x"],"token_delta":0,"tuple_leak_window_count":40},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"nfkc":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"nfkc_cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"raw":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"ws_collapse":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"}}},{"arm_id":"cycle7_combined","domain":"conversational_prose","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":53,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":46,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":7,"measurement_hash":"aeaf328f84e96fddc16b96dd6850fe96a4ec58760c6bfb0fb9848992c98920ef","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","expand-you-are","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","expand-that-is-not-x","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":12,"tuple_leak_window_count":8},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"nfkc":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"nfkc_cf_strip":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"raw":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"ws_collapse":{"detected":false,"score":0.4962785752259436,"text_hash":"9d238528d87390bd78adb434049836f4b4c913567ff043a539c90e5b4128c06d"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.4962785752259436,"text_hash":"9d238528d87390bd78adb434049836f4b4c913567ff043a539c90e5b4128c06d"}}},{"arm_id":"cycle6_spacing_b14","domain":"conversational_prose","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"fa9c76c6ddb778141a4962a77e8002c1413ca7289d88c89177fe9d861683a4b5","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc_cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"raw":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"ws_collapse":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"}}},{"arm_id":"cycle7_durable","domain":"conversational_prose","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"a815a53516b54ea1011be7ad4d6e8fc09b998a1bd7770f4c7171fad6431e4c4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"nfkc":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"raw":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"}}},{"arm_id":"cycle7_combined","domain":"conversational_prose","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"3868bda9994b005506d125b7a70b2110f37b3b27cdb51c088ee19cd636b20c71","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc_cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"raw":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"ws_collapse":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"}}},{"arm_id":"cycle6_spacing_b14","domain":"structured_instructional","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"c89d29f8c4ff8d8b907f91b255343f24aebb113f6f990e9dba00911566095e4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc_cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"raw":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle7_durable","domain":"structured_instructional","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"d8e63ad3f8cddab8110faa1c91ecd15fc2161d622d41d3a655252553eb70c92a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"nfkc":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"raw":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle7_combined","domain":"structured_instructional","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"9483f967307734bda93606f643cfad5f2ad1632797995b994ce16324a0dbfffc","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc_cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"raw":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle6_spacing_b14","domain":"structured_instructional","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":44,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":13,"measurement_hash":"4fa592661bb841d8a40f7403a54f3c091faf666d726a680b733527f54d98a981","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":29,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma"],"token_delta":14,"tuple_leak_window_count":13},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"nfkc":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"nfkc_cf_strip":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"raw":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"ws_collapse":{"detected":true,"score":0.653030303030303,"text_hash":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.653030303030303,"text_hash":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"}}},{"arm_id":"cycle7_durable","domain":"structured_instructional","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":3,"candidate_exhausted":true,"changed_token_positions":18,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":44,"measurement_hash":"79aae0d2f1050d53e025cf30201df0f5f5cc6604cfd556a2384002c1df777cb9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":3,"selected_rule_ids":["cycle7-expand-its-like","expand-cannot","cycle7-expand-i-have"],"token_delta":-1,"tuple_leak_window_count":44},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"nfkc":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"nfkc_cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"raw":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"ws_collapse":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"}}},{"arm_id":"cycle7_combined","domain":"structured_instructional","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":47,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":9,"measurement_hash":"26ce88d91c96c2cd4c2f98a4418a6b9d82afe975d7d9e9f897d1734cfb75c4af","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","cycle7-expand-its-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","expand-cannot","surface-space-after-any-word","cycle7-expand-i-have","surface-space-after-comma"],"token_delta":10,"tuple_leak_window_count":9},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"nfkc":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"nfkc_cf_strip":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"raw":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"ws_collapse":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"}}}],"secret_access_used_for_selection":false,"seed_base":810000,"summaries":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5254488114855487,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6426760741438237},"cycle7_combined":{"arm_id":"cycle7_combined","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5239653475020847,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6295426209679976},"cycle7_durable":{"arm_id":"cycle7_durable","median_raw_watermarked_score":0.6286616161616162,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":4,"raw_watermarked_mean_score":0.6275224189477956,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.6634137990070192,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6275224189477956}},"threshold":0.5570987654320988} diff --git a/evidence/cycle7-stage-a-2026-08-25/fixture-geometry.json b/evidence/cycle7-stage-a-2026-08-25/fixture-geometry.json new file mode 100644 index 00000000..b705cbf7 --- /dev/null +++ b/evidence/cycle7-stage-a-2026-08-25/fixture-geometry.json @@ -0,0 +1 @@ +{"algorithm_version":"cycle7-stage-a-fixture-compare-v1","arm_ids":["cycle6_spacing_b14","cycle7_durable","cycle7_combined"],"artifact_hash":"f76f56efcdad95c38fce42289cfa6a2689bf99436e37fd63d237db9b144c80ad","budget":14,"detector":{"available":true,"combined":{"arm_id":"cycle7_combined","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5239653475020847,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6295426209679976},"cycle6_spacing":{"arm_id":"cycle6_spacing_b14","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5254488114855487,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6426760741438237},"durable":{"arm_id":"cycle7_durable","median_raw_watermarked_score":0.6286616161616162,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":4,"raw_watermarked_mean_score":0.6275224189477956,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.6634137990070192,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6275224189477956}},"rows":[{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":77,"candidate_exhausted":false,"changed_token_positions":473,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":408,"collapsed_token_delta":-3,"collapsed_tuple_leak_window_count":411,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":356,"measurement_hash":"67c9eb7ce713eebb51900bbd991e28abdd07873491114baf405a410c0ad13fe6","protected_blocked_count":4,"quote_blocked_count":1,"reachable_unselected_static_cover":51,"root_window_count":426,"selected_count":14,"selected_rule_ids":["contract-are-not","contract-do-not","surface-space-after-any-word","contract-cannot","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-before-sentence","surface-space-before-sentence"],"token_delta":8,"tuple_leak_window_count":378},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":81,"candidate_exhausted":false,"changed_token_positions":418,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":396,"collapsed_token_delta":-7,"collapsed_tuple_leak_window_count":399,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":353,"measurement_hash":"8b9061d597e0f732a02c050de3620e93f2ec05acd9dcfc5d97dc1082cb7d3db7","protected_blocked_count":4,"quote_blocked_count":0,"reachable_unselected_static_cover":51,"root_window_count":426,"selected_count":14,"selected_rule_ids":["contract-are-not","contract-do-not","surface-space-after-any-word","contract-cannot","cycle7-contract-they-have","surface-space-after-any-word","cycle7-ortho-amongst-among","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-before-sentence","surface-space-before-sentence"],"token_delta":2,"tuple_leak_window_count":372},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":9,"candidate_exhausted":true,"changed_token_positions":475,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":381,"collapsed_token_delta":-10,"collapsed_tuple_leak_window_count":382,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":381,"measurement_hash":"675f24e72b0b7dd4ac7a0b5f3c25b55c8122e28e3dca8fff20e5a2a2abd999fe","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":426,"selected_count":8,"selected_rule_ids":["contract-are-not","contract-do-not","contract-cannot","contract-i-am","cycle7-contract-they-have","cycle7-ortho-towards-toward","cycle7-ortho-amongst-among","contract-do-not"],"token_delta":-10,"tuple_leak_window_count":382}},"source_sample_id":"contraction-rich","source_text_hash":"75445d3d08c371e274c0c6a1fb9b98e374e0eb60e5cf373a7921499771d41e91"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":18,"candidate_exhausted":false,"changed_token_positions":137,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":132,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":132,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":68,"measurement_hash":"50dbcb37bbc266eb6f5cccff5f597629f059f4730eb84c5abb545baf9aefe150","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":2,"root_window_count":132,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":94},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":18,"candidate_exhausted":false,"changed_token_positions":137,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":132,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":132,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":68,"measurement_hash":"e765da5b67fc654decea5cf2a1ebda2daa2addabd420210e3abdb3ee1d39166c","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":2,"root_window_count":132,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":94},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":132,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":132,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":132,"measurement_hash":"cb8747086fa20595a37904be0cb07aff1d90e82557a8b5086e28253e750a1460","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":132,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":132}},"source_sample_id":"contraction-sparse","source_text_hash":"ef1e423456e6e8bf14408dd5e80ea6a491938c17541dd00a3fa90695ebecbed3"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":false,"candidate_count":10,"candidate_exhausted":true,"changed_token_positions":61,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":51,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":51,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":13,"measurement_hash":"20550c7fb44d950468245109b89a29d7185a2b5703fae7fafeed03c169277809","protected_blocked_count":0,"quote_blocked_count":3,"reachable_unselected_static_cover":0,"root_window_count":51,"selected_count":10,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":10,"tuple_leak_window_count":23},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":false,"candidate_count":13,"candidate_exhausted":true,"changed_token_positions":56,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":-2,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":17,"measurement_hash":"f23745fd66429f0f64e1388eec798c49f8bb13b47bf2cd2d7c5bd9228efdbec8","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":51,"selected_count":8,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-comma","contract-they-are","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","contract-do-not"],"token_delta":4,"tuple_leak_window_count":24},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":3,"candidate_exhausted":true,"changed_token_positions":39,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":-2,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":40,"measurement_hash":"c2681e52e67c056c661046bc533cf823801f91c6b552e8265681af76aa26a31b","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":51,"selected_count":2,"selected_rule_ids":["contract-they-are","contract-do-not"],"token_delta":-2,"tuple_leak_window_count":40}},"source_sample_id":"quote-interior","source_text_hash":"9435a82cc500b4ec86c09b20965a484ebea69f0ae0603cdf0a057a9b632c93ba"}]} diff --git a/fuckmark/cycle7/__init__.py b/fuckmark/cycle7/__init__.py new file mode 100644 index 00000000..19cf6f7a --- /dev/null +++ b/fuckmark/cycle7/__init__.py @@ -0,0 +1,37 @@ +from __future__ import annotations + +from .ledger import ( + CYCLE7_CONFIRMATION_RESERVED_SEED_BASES, + CYCLE7_EXPLORATORY_SEED_BASE, + CYCLE7_LEDGER_VERSION, + CYCLE7_VALIDATION_SEED_BASE, + SPENT_CONFIRMATION_SEED_BASES, + SPENT_DEVELOPMENT_SEED_BASES, + assert_development_seed, +) +from .registry import ( + cycle7_combined_transform_registry, + cycle7_durable_transform_registry, +) +from .whitespace_collapse import ( + CYCLE7_SANITIZER_VARIANT_IDS, + WHITESPACE_COLLAPSE_VERSION, + collapse_horizontal_ascii_whitespace, + sanitize_cycle7_variant, +) + +__all__ = [ + "CYCLE7_CONFIRMATION_RESERVED_SEED_BASES", + "CYCLE7_EXPLORATORY_SEED_BASE", + "CYCLE7_LEDGER_VERSION", + "CYCLE7_SANITIZER_VARIANT_IDS", + "CYCLE7_VALIDATION_SEED_BASE", + "SPENT_CONFIRMATION_SEED_BASES", + "SPENT_DEVELOPMENT_SEED_BASES", + "WHITESPACE_COLLAPSE_VERSION", + "assert_development_seed", + "collapse_horizontal_ascii_whitespace", + "cycle7_combined_transform_registry", + "cycle7_durable_transform_registry", + "sanitize_cycle7_variant", +] diff --git a/fuckmark/cycle7/compare.py b/fuckmark/cycle7/compare.py new file mode 100644 index 00000000..d593a34d --- /dev/null +++ b/fuckmark/cycle7/compare.py @@ -0,0 +1,103 @@ +from __future__ import annotations + +from typing import Any + +from ..experiments.cycle6_confirmation import CYCLE6_BUDGET +from ..hashing import sha256_json, sha256_text +from ..transforms import quote_safe_zrd_transform_registry +from .fixtures import fixture_samples +from .instrumentation import Cycle7ArmMeasurement, measure_arm +from .registry import cycle7_combined_transform_registry, cycle7_durable_transform_registry + + +CYCLE7_STAGE_A_COMPARE_VERSION = "cycle7-stage-a-fixture-compare-v1" +CYCLE6_SPACING_ARM_ID = "cycle6_spacing_b14" +CYCLE7_DURABLE_ARM_ID = "cycle7_durable" +CYCLE7_COMBINED_ARM_ID = "cycle7_combined" + + +def _registries(): + return { + CYCLE6_SPACING_ARM_ID: quote_safe_zrd_transform_registry(), + CYCLE7_DURABLE_ARM_ID: cycle7_durable_transform_registry(), + CYCLE7_COMBINED_ARM_ID: cycle7_combined_transform_registry(), + } + + +def compare_arms_on_text( + *, + source_sample_id: str, + source_text: str, + tokenizer: Any, + tokenizer_identity_hash: str, + budget: int = CYCLE6_BUDGET, +) -> dict[str, Cycle7ArmMeasurement]: + return { + arm_id: measure_arm( + arm_id=arm_id, + source_sample_id=source_sample_id, + source_text=source_text, + registry=registry, + tokenizer=tokenizer, + tokenizer_identity_hash=tokenizer_identity_hash, + budget=budget, + ) + for arm_id, registry in _registries().items() + } + + +def summarize_arm(measurement: Cycle7ArmMeasurement) -> dict[str, object]: + return { + "arm_id": measurement.arm_id, + "candidate_count": measurement.candidate_count, + "selected_count": measurement.selected_count, + "selected_rule_ids": measurement.selected_rule_ids, + "budget_exhausted": measurement.budget_exhausted, + "candidate_exhausted": measurement.candidate_exhausted, + "protected_blocked_count": measurement.protected_blocked_count, + "quote_blocked_count": measurement.quote_blocked_count, + "token_delta": measurement.token_delta, + "collapsed_token_delta": measurement.collapsed_token_delta, + "changed_token_positions": measurement.changed_token_positions, + "root_window_count": measurement.root_window_count, + "intact_window_count": measurement.intact_window_count, + "tuple_leak_window_count": measurement.tuple_leak_window_count, + "closure_free": measurement.closure_free, + "collapsed_intact_window_count": measurement.collapsed_intact_window_count, + "collapsed_tuple_leak_window_count": measurement.collapsed_tuple_leak_window_count, + "collapsed_closure_free": measurement.collapsed_closure_free, + "collapsed_equals_collapsed_source": measurement.collapsed_equals_collapsed_source, + "reachable_unselected_static_cover": measurement.reachable_unselected_static_cover, + "failure_classes": measurement.failure_classes, + "measurement_hash": measurement.measurement_hash, + } + + +def run_fixture_stage_a( + tokenizer: Any, + tokenizer_identity_hash: str | None = None, +) -> dict[str, object]: + if tokenizer_identity_hash is None: + tokenizer_identity_hash = sha256_text("cycle7-fixture-tokenizer") + rows = [] + for sample_id, text in fixture_samples(): + arms = compare_arms_on_text( + source_sample_id=sample_id, + source_text=text, + tokenizer=tokenizer, + tokenizer_identity_hash=tokenizer_identity_hash, + ) + rows.append( + { + "source_sample_id": sample_id, + "source_text_hash": sha256_text(text), + "arms": {arm_id: summarize_arm(measurement) for arm_id, measurement in arms.items()}, + } + ) + payload = { + "algorithm_version": CYCLE7_STAGE_A_COMPARE_VERSION, + "budget": CYCLE6_BUDGET, + "arm_ids": (CYCLE6_SPACING_ARM_ID, CYCLE7_DURABLE_ARM_ID, CYCLE7_COMBINED_ARM_ID), + "rows": tuple(rows), + } + return {**payload, "artifact_hash": sha256_json(payload)} diff --git a/fuckmark/cycle7/decision.py b/fuckmark/cycle7/decision.py new file mode 100644 index 00000000..7b9427cc --- /dev/null +++ b/fuckmark/cycle7/decision.py @@ -0,0 +1,137 @@ +from __future__ import annotations + +from collections.abc import Mapping + +from .compare import ( + CYCLE6_SPACING_ARM_ID, + CYCLE7_COMBINED_ARM_ID, + CYCLE7_DURABLE_ARM_ID, +) + + +CYCLE7_DECISION_VERSION = "cycle7-stage-a-decision-v1" +PROMISING_DEVELOPMENT = "PROMISING_DEVELOPMENT" +REJECTED = "REJECTED" +INSUFFICIENT_EVIDENCE = "INSUFFICIENT_EVIDENCE" + + +def _arm(row: Mapping[str, object], arm_id: str) -> Mapping[str, object]: + arms = row["arms"] + if not isinstance(arms, Mapping): + raise TypeError("row arms must be a mapping") + value = arms[arm_id] + if not isinstance(value, Mapping): + raise TypeError("arm summary must be a mapping") + return value + + +def classify_fixture_stage_a(report: Mapping[str, object]) -> dict[str, object]: + rows = report.get("rows") + if not isinstance(rows, (list, tuple)) or not rows: + raise ValueError("Stage A report must contain rows") + rich = next(row for row in rows if row["source_sample_id"] == "contraction-rich") + sparse = next(row for row in rows if row["source_sample_id"] == "contraction-sparse") + spacing_rich = _arm(rich, CYCLE6_SPACING_ARM_ID) + durable_rich = _arm(rich, CYCLE7_DURABLE_ARM_ID) + combined_rich = _arm(rich, CYCLE7_COMBINED_ARM_ID) + durable_sparse = _arm(sparse, CYCLE7_DURABLE_ARM_ID) + + quote = next(row for row in rows if row["source_sample_id"] == "quote-interior") + spacing_quote = _arm(quote, CYCLE6_SPACING_ARM_ID) + durable_quote = _arm(quote, CYCLE7_DURABLE_ARM_ID) + collapse_survives = durable_rich["collapsed_equals_collapsed_source"] is False + spacing_undone = spacing_rich["collapsed_equals_collapsed_source"] is True + durable_has_ops = int(durable_rich["selected_count"]) > 0 + sparse_density = int(durable_sparse["candidate_count"]) + combined_has_ops = int(combined_rich["selected_count"]) > 0 + + reasons: list[str] = [] + if durable_has_ops and collapse_survives: + reasons.append("durable edits remain after ASCII whitespace collapse on contraction-rich text") + if spacing_undone and int(spacing_rich["selected_count"]) > 0: + reasons.append("Cycle 6 spacing edits are removed by the same collapse sanitizer") + if ( + spacing_quote["collapsed_equals_collapsed_source"] is True + and durable_quote["collapsed_equals_collapsed_source"] is False + and int(durable_quote["selected_count"]) > 0 + ): + reasons.append( + "inside quotes, Cycle 6 spacing collapses away while Cycle 7 durable edits remain" + ) + if sparse_density == 0: + reasons.append("durable candidate density is zero on contraction-sparse text") + if combined_has_ops: + reasons.append("combined durable+spacing still enumerates a non-empty candidate set") + + detector = report.get("detector") + detector_signal = None + if isinstance(detector, Mapping) and detector.get("available") is True: + durable_collapse_detected = int(detector["durable"]["ws_collapse_watermarked_detected"]) + spacing_collapse_detected = int(detector["cycle6_spacing"]["ws_collapse_watermarked_detected"]) + durable_raw_detected = int(detector["durable"]["raw_watermarked_detected"]) + spacing_raw_detected = int(detector["cycle6_spacing"]["raw_watermarked_detected"]) + n_watermarked = int(detector["durable"]["watermarked_row_count"]) + detector_signal = { + "durable_raw_detected": durable_raw_detected, + "spacing_raw_detected": spacing_raw_detected, + "durable_ws_collapse_detected": durable_collapse_detected, + "spacing_ws_collapse_detected": spacing_collapse_detected, + "watermarked_row_count": n_watermarked, + } + if spacing_raw_detected < n_watermarked and spacing_collapse_detected == n_watermarked: + reasons.append( + "Cycle 6 spacing reduced raw detections and those detections returned after whitespace collapse" + ) + if durable_raw_detected == n_watermarked: + reasons.append("durable family did not reduce raw detections on the Stage A corpus") + elif durable_collapse_detected < spacing_collapse_detected: + reasons.append("durable arm has fewer ws_collapse detections than Cycle 6 spacing on the same corpus") + elif durable_collapse_detected == spacing_collapse_detected: + reasons.append("durable and Cycle 6 spacing ws_collapse detection counts are tied") + else: + reasons.append("durable arm did not beat Cycle 6 spacing on ws_collapse detections") + + if not durable_has_ops: + decision = REJECTED + reasons.append("durable family selected no operations on contraction-rich text") + elif not collapse_survives: + decision = REJECTED + reasons.append("durable family did not survive whitespace collapse") + elif detector is None: + decision = PROMISING_DEVELOPMENT + reasons.append("geometry-only Stage A; detector scoring was not attached") + elif isinstance(detector, Mapping) and detector.get("available") is not True: + decision = PROMISING_DEVELOPMENT + reasons.append("geometry collapse-survival demonstrated; detector scoring unavailable") + else: + durable_collapse_detected = int(detector["durable"]["ws_collapse_watermarked_detected"]) + spacing_collapse_detected = int(detector["cycle6_spacing"]["ws_collapse_watermarked_detected"]) + durable_raw_detected = int(detector["durable"]["raw_watermarked_detected"]) + n_watermarked = int(detector["durable"]["watermarked_row_count"]) + if durable_collapse_detected < spacing_collapse_detected: + decision = PROMISING_DEVELOPMENT + reasons.append("this tiny corpus is not a formal confirmation") + elif durable_raw_detected == n_watermarked: + decision = INSUFFICIENT_EVIDENCE + reasons.append( + "durable collapse-survival is real on fixtures, but density is too low to replace Cycle 6 spacing" + ) + elif durable_has_ops and collapse_survives: + decision = PROMISING_DEVELOPMENT + reasons.append( + "collapse-survival is demonstrated, but this tiny corpus is not a formal confirmation" + ) + else: + decision = INSUFFICIENT_EVIDENCE + + payload = { + "algorithm_version": CYCLE7_DECISION_VERSION, + "decision": decision, + "reasons": tuple(reasons), + "detector_signal": detector_signal, + "notes": ( + "This is a development classification, not a Cycle 7 formal confirmation. " + "Confirmation seeds 830000/840000/850000 were not inspected." + ), + } + return payload diff --git a/fuckmark/cycle7/durable_rules.py b/fuckmark/cycle7/durable_rules.py new file mode 100644 index 00000000..5a780042 --- /dev/null +++ b/fuckmark/cycle7/durable_rules.py @@ -0,0 +1,337 @@ +from __future__ import annotations + +from dataclasses import dataclass + +from .._validation import require_clean_string, require_sha256 +from ..hashing import sha256_json +from ..transforms.contractions import ( + development_forward_contraction_rules, + reverse_contraction_rules, + zrd_forward_contraction_extension_rules, + zrd_reverse_contraction_extension_rules, +) +from ..transforms.lexical_rules import development_lexical_rules +from ..transforms.rules import LiteralTransformRule +from ..transforms.schema import TransformFamily, TransformTier +from ..transforms.syntax_rules import development_syntax_rules + + +CYCLE7_DURABLE_RULE_CATALOG_VERSION = "cycle7-durable-rule-catalog-v1" + + +@dataclass(frozen=True, slots=True) +class Cycle7DurableRewrite: + rule_id: str + inverse_rule_id: str + family: TransformFamily + source: str + replacement: str + inverse_group_id: str + whitespace_collapse_survives: bool + ambiguous: bool + notes: str + metadata_hash: str + + def __post_init__(self) -> None: + for name in ("rule_id", "inverse_rule_id", "inverse_group_id", "notes"): + require_clean_string(name, getattr(self, name)) + if not isinstance(self.source, str) or not self.source: + raise ValueError("source must be non-empty") + if not isinstance(self.replacement, str) or not self.replacement: + raise ValueError("replacement must be non-empty") + if self.source.casefold() == self.replacement.casefold(): + raise ValueError("source and replacement must differ") + if self.family not in (TransformFamily.CONTRACTION, TransformFamily.ORTHOGRAPHY): + raise ValueError("Cycle 7 durable rewrites must be contraction or orthography") + if self.ambiguous: + raise ValueError("ambiguous Cycle 7 rewrites are not admitted") + if not self.whitespace_collapse_survives: + raise ValueError("Cycle 7 durable rewrites must survive whitespace collapse") + require_sha256("metadata_hash", self.metadata_hash) + if self.metadata_hash != sha256_json(self.payload()): + raise ValueError("metadata_hash does not match Cycle 7 durable rewrite") + + def payload(self) -> dict[str, object]: + return { + "algorithm_version": CYCLE7_DURABLE_RULE_CATALOG_VERSION, + "rule_id": self.rule_id, + "inverse_rule_id": self.inverse_rule_id, + "family": self.family.value, + "source": self.source, + "replacement": self.replacement, + "inverse_group_id": self.inverse_group_id, + "whitespace_collapse_survives": self.whitespace_collapse_survives, + "ambiguous": self.ambiguous, + "notes": self.notes, + } + + @classmethod + def create( + cls, + *, + rule_id: str, + inverse_rule_id: str, + family: TransformFamily, + source: str, + replacement: str, + inverse_group_id: str, + notes: str, + ) -> Cycle7DurableRewrite: + payload = { + "algorithm_version": CYCLE7_DURABLE_RULE_CATALOG_VERSION, + "rule_id": rule_id, + "inverse_rule_id": inverse_rule_id, + "family": family.value, + "source": source, + "replacement": replacement, + "inverse_group_id": inverse_group_id, + "whitespace_collapse_survives": True, + "ambiguous": False, + "notes": notes, + } + return cls( + rule_id=rule_id, + inverse_rule_id=inverse_rule_id, + family=family, + source=source, + replacement=replacement, + inverse_group_id=inverse_group_id, + whitespace_collapse_survives=True, + ambiguous=False, + notes=notes, + metadata_hash=sha256_json(payload), + ) + + +def cycle7_new_contraction_metadata() -> tuple[Cycle7DurableRewrite, ...]: + rows = ( + ( + "cycle7-contract-i-have", + "cycle7-expand-i-have", + "I have", + "I've", + "cycle7-contraction-i-have", + "Unambiguous auxiliary have. Inverse restores 'I have'.", + ), + ( + "cycle7-contract-you-have", + "cycle7-expand-you-have", + "you have", + "you've", + "cycle7-contraction-you-have", + "Unambiguous auxiliary have.", + ), + ( + "cycle7-contract-we-have", + "cycle7-expand-we-have", + "we have", + "we've", + "cycle7-contraction-we-have", + "Unambiguous auxiliary have.", + ), + ( + "cycle7-contract-they-have", + "cycle7-expand-they-have", + "they have", + "they've", + "cycle7-contraction-they-have", + "Unambiguous auxiliary have.", + ), + ) + return tuple( + Cycle7DurableRewrite.create( + rule_id=rule_id, + inverse_rule_id=inverse_rule_id, + family=TransformFamily.CONTRACTION, + source=source, + replacement=replacement, + inverse_group_id=group_id, + notes=notes, + ) + for rule_id, inverse_rule_id, source, replacement, group_id, notes in rows + ) + + +def cycle7_bounded_copula_metadata() -> tuple[Cycle7DurableRewrite, ...]: + rows = ( + ( + "cycle7-expand-its-like", + "cycle7-contract-its-like", + "it's like", + "it is like", + "cycle7-copula-its-like", + "it's+like is it is. Bare it's and it's not are excluded: it's not trips hard-invariant negation identity, and it's is otherwise is/has-ambiguous.", + ), + ( + "cycle7-expand-its-important", + "cycle7-contract-its-important", + "it's important", + "it is important", + "cycle7-copula-its-important", + "it's+important is it is.", + ), + ( + "cycle7-expand-its-a", + "cycle7-contract-its-a", + "it's a", + "it is a", + "cycle7-copula-its-a", + "it's+a is it is. Does not match it's been.", + ), + ( + "cycle7-expand-its-an", + "cycle7-contract-its-an", + "it's an", + "it is an", + "cycle7-copula-its-an", + "it's+an is it is.", + ), + ( + "cycle7-expand-its-the", + "cycle7-contract-its-the", + "it's the", + "it is the", + "cycle7-copula-its-the", + "it's+the is it is.", + ), + ) + return tuple( + Cycle7DurableRewrite.create( + rule_id=rule_id, + inverse_rule_id=inverse_rule_id, + family=TransformFamily.CONTRACTION, + source=source, + replacement=replacement, + inverse_group_id=group_id, + notes=notes, + ) + for rule_id, inverse_rule_id, source, replacement, group_id, notes in rows + ) + + +def cycle7_orthography_metadata() -> tuple[Cycle7DurableRewrite, ...]: + rows = ( + ( + "cycle7-ortho-towards-toward", + "cycle7-ortho-toward-towards", + "towards", + "toward", + "cycle7-ortho-toward", + "UK/US prepositional variant. Token boundary change survives space collapse.", + ), + ( + "cycle7-ortho-toward-towards", + "cycle7-ortho-towards-toward", + "toward", + "towards", + "cycle7-ortho-toward", + "Inverse of towards/toward. Adjective 'toward' is rare; whole-word still required.", + ), + ( + "cycle7-ortho-amongst-among", + "cycle7-ortho-among-amongst", + "amongst", + "among", + "cycle7-ortho-among", + "UK/US prepositional variant.", + ), + ( + "cycle7-ortho-among-amongst", + "cycle7-ortho-amongst-among", + "among", + "amongst", + "cycle7-ortho-among", + "Inverse of amongst/among.", + ), + ) + return tuple( + Cycle7DurableRewrite.create( + rule_id=rule_id, + inverse_rule_id=inverse_rule_id, + family=TransformFamily.ORTHOGRAPHY, + source=source, + replacement=replacement, + inverse_group_id=group_id, + notes=notes, + ) + for rule_id, inverse_rule_id, source, replacement, group_id, notes in rows + ) + + +def _rules_from_metadata(rows: tuple[Cycle7DurableRewrite, ...]) -> tuple[LiteralTransformRule, ...]: + return tuple( + LiteralTransformRule.create( + rule_id=row.rule_id, + version="v1", + family=row.family, + tier=TransformTier.SURFACE, + source=row.source, + replacement=row.replacement, + ) + for row in rows + ) + + +def cycle7_new_contraction_rules() -> tuple[LiteralTransformRule, ...]: + forward = _rules_from_metadata(cycle7_new_contraction_metadata()) + reverse_meta = tuple( + Cycle7DurableRewrite.create( + rule_id=row.inverse_rule_id, + inverse_rule_id=row.rule_id, + family=row.family, + source=row.replacement, + replacement=row.source, + inverse_group_id=row.inverse_group_id, + notes=f"Inverse of {row.rule_id}.", + ) + for row in cycle7_new_contraction_metadata() + ) + return (*forward, *_rules_from_metadata(reverse_meta)) + + +def cycle7_bounded_copula_rules() -> tuple[LiteralTransformRule, ...]: + forward = _rules_from_metadata(cycle7_bounded_copula_metadata()) + reverse_meta = tuple( + Cycle7DurableRewrite.create( + rule_id=row.inverse_rule_id, + inverse_rule_id=row.rule_id, + family=row.family, + source=row.replacement, + replacement=row.source, + inverse_group_id=row.inverse_group_id, + notes=f"Inverse of {row.rule_id}.", + ) + for row in cycle7_bounded_copula_metadata() + ) + return (*forward, *_rules_from_metadata(reverse_meta)) + + +def cycle7_orthography_rules() -> tuple[LiteralTransformRule, ...]: + return _rules_from_metadata(cycle7_orthography_metadata()) + + +def cycle7_durable_rules() -> tuple[object, ...]: + return ( + *development_forward_contraction_rules(), + *reverse_contraction_rules(), + *zrd_forward_contraction_extension_rules(), + *zrd_reverse_contraction_extension_rules(), + *cycle7_new_contraction_rules(), + *cycle7_bounded_copula_rules(), + *cycle7_orthography_rules(), + *development_lexical_rules(), + *development_syntax_rules(), + ) + + +def rejected_ambiguous_contraction_examples() -> tuple[tuple[str, str], ...]: + return ( + ("he would", "he'd"), + ("she would", "she'd"), + ("I would", "I'd"), + ("it is", "it's"), + ("that is", "that's"), + ("let us", "let's"), + ("it's not", "it is not"), + ("it's been", "it is been"), + ) diff --git a/fuckmark/cycle7/fixtures.py b/fuckmark/cycle7/fixtures.py new file mode 100644 index 00000000..dd1e51b5 --- /dev/null +++ b/fuckmark/cycle7/fixtures.py @@ -0,0 +1,54 @@ +from __future__ import annotations + +from ..transforms.schema import CandidateRejectionReason + + +CYCLE7_FIXTURE_SET_ID = "cycle7-stage-a-fixtures-v1" + +CONTRACTION_RICH = ( + "They are not ready, and we do not stop when the system is in use. " + "You cannot wait, and I am sure they have time. The path towards the lab " + "is amongst the remaining options. She said, \"Do not quote this change.\" " + "See https://example.com/path and 12.5% on 2024-08-25. Run `code_sample` " + "and pass --verbose. Contact lab@example.com. Cite [1] and (Smith, 2024). " + "Hash deadbeefcafebabe0123456789abcdef0123456789abcdef0123456789abcdef. " + "Path /usr/bin/true and C:\\Windows\\System32\\drivers. Identifier CamelCaseThing." +) + +CONTRACTION_SPARSE = ( + "Measurement protocol remains fixed across independent replications of the " + "same experimental apparatus. Calibration uses a held-out population." +) + +QUOTE_INTERIOR = 'He answered, "They are not finished and we do not agree."' + +AMBIGUOUS_NEGATIVES = ( + "He'd already left before the meeting started.", + "It's time for the measurement to begin.", + "Let us know when the run completes.", +) + +PROTECTED_CASES = ( + ("url", "Visit https://example.com/do-not-touch and continue."), + ("number", "The value 12 is exact and we do not round it."), + ("date", "On 2024-08-25 we do not change the timestamp."), + ("code", "Run `do not` as a literal command token."), + ("identifier", "Keep CamelCaseThing intact while they are waiting."), + ("posix_path", "The binary /usr/bin/true cannot be rewritten."), + ("windows_path", "File C:\\Windows\\System32\\drivers cannot be rewritten."), + ("hash", "Digest deadbeefcafebabe0123456789abcdef0123456789abcdef0123456789abcdef is fixed."), + ("citation", "Prior work [1] cannot be silently altered."), + ("email", "Write lab@example.com if they are delayed."), +) + + +def fixture_samples() -> tuple[tuple[str, str], ...]: + return ( + ("contraction-rich", CONTRACTION_RICH), + ("contraction-sparse", CONTRACTION_SPARSE), + ("quote-interior", QUOTE_INTERIOR), + ) + + +def rejection_reason_values() -> tuple[str, ...]: + return tuple(reason.value for reason in CandidateRejectionReason) diff --git a/fuckmark/cycle7/instrumentation.py b/fuckmark/cycle7/instrumentation.py new file mode 100644 index 00000000..12827715 --- /dev/null +++ b/fuckmark/cycle7/instrumentation.py @@ -0,0 +1,286 @@ +from __future__ import annotations + +from collections import Counter +from dataclasses import dataclass, replace +from typing import Any + +from .._validation import require_clean_string, require_int, require_sha256 +from ..experiments.cover_greedy_v4 import schedule_cover_greedy_v4 +from ..geometry.counterfactual import CounterfactualGeometryEngine, GeometryConfig +from ..geometry.repetition import PublicRepetitionGeometry +from ..geometry.tuple_closure import compute_tuple_closure +from ..hashing import sha256_json, sha256_text +from ..transforms.registry import TransformRegistry +from ..transforms.schema import CandidateRejectionReason +from .whitespace_collapse import collapse_horizontal_ascii_whitespace + + +CYCLE7_INSTRUMENTATION_VERSION = "cycle7-arm-instrumentation-v1" + + +@dataclass(frozen=True, slots=True) +class Cycle7ArmMeasurement: + algorithm_version: str + arm_id: str + source_sample_id: str + source_text_hash: str + candidate_count: int + selected_count: int + selected_rule_ids: tuple[str, ...] + budget: int + budget_exhausted: bool + candidate_exhausted: bool + rejection_counts: dict[str, int] + protected_blocked_count: int + quote_blocked_count: int + conflict_excluded_count: int + unselected_count: int + token_count_source: int + token_count_transformed: int + token_count_collapsed: int + token_delta: int + collapsed_token_delta: int + changed_token_positions: int + root_window_count: int + intact_window_count: int + tuple_leak_window_count: int + closure_free: bool + collapsed_intact_window_count: int + collapsed_tuple_leak_window_count: int + collapsed_closure_free: bool + collapsed_equals_collapsed_source: bool + reachable_unselected_static_cover: int + failure_classes: tuple[str, ...] + transformed_text: str + collapsed_text: str + measurement_hash: str + + def payload(self) -> dict[str, object]: + return { + "algorithm_version": self.algorithm_version, + "arm_id": self.arm_id, + "source_sample_id": self.source_sample_id, + "source_text_hash": self.source_text_hash, + "candidate_count": self.candidate_count, + "selected_count": self.selected_count, + "selected_rule_ids": self.selected_rule_ids, + "budget": self.budget, + "budget_exhausted": self.budget_exhausted, + "candidate_exhausted": self.candidate_exhausted, + "rejection_counts": self.rejection_counts, + "protected_blocked_count": self.protected_blocked_count, + "quote_blocked_count": self.quote_blocked_count, + "conflict_excluded_count": self.conflict_excluded_count, + "unselected_count": self.unselected_count, + "token_count_source": self.token_count_source, + "token_count_transformed": self.token_count_transformed, + "token_count_collapsed": self.token_count_collapsed, + "token_delta": self.token_delta, + "collapsed_token_delta": self.collapsed_token_delta, + "changed_token_positions": self.changed_token_positions, + "root_window_count": self.root_window_count, + "intact_window_count": self.intact_window_count, + "tuple_leak_window_count": self.tuple_leak_window_count, + "closure_free": self.closure_free, + "collapsed_intact_window_count": self.collapsed_intact_window_count, + "collapsed_tuple_leak_window_count": self.collapsed_tuple_leak_window_count, + "collapsed_closure_free": self.collapsed_closure_free, + "collapsed_equals_collapsed_source": self.collapsed_equals_collapsed_source, + "reachable_unselected_static_cover": self.reachable_unselected_static_cover, + "failure_classes": self.failure_classes, + "transformed_text_hash": sha256_text(self.transformed_text), + "collapsed_text_hash": sha256_text(self.collapsed_text), + } + + +def _encode(tokenizer: Any, text: str) -> tuple[int, ...]: + encoded = tokenizer(text, add_special_tokens=False) + ids = encoded["input_ids"] + if ids and isinstance(ids[0], list): + ids = ids[0] + return tuple(int(value) for value in ids) + + +def _changed_positions(left: tuple[int, ...], right: tuple[int, ...]) -> int: + shared = min(len(left), len(right)) + changed = sum(left[index] != right[index] for index in range(shared)) + return changed + abs(len(left) - len(right)) + + +def _rejection_counts(enumeration) -> dict[str, int]: + counts: Counter[str] = Counter(rejection.reason.value for rejection in enumeration.rejections) + return {reason.value: int(counts.get(reason.value, 0)) for reason in CandidateRejectionReason} + + +def _failure_classes(measurement_fields: dict[str, Any]) -> tuple[str, ...]: + labels: list[str] = [] + if measurement_fields["candidate_count"] == 0: + labels.append("insufficient_transform_density") + elif measurement_fields["selected_count"] == 0: + labels.append("insufficient_editable_reachability") + if measurement_fields["budget_exhausted"]: + labels.append("budget_exhaustion") + if measurement_fields["candidate_exhausted"]: + labels.append("candidate_exhaustion") + if measurement_fields["protected_blocked_count"] > 0 and measurement_fields["selected_count"] == 0: + labels.append("protected_spans") + if measurement_fields["quote_blocked_count"] > 0 and measurement_fields["selected_count"] == 0: + labels.append("quote_boundaries") + if ( + measurement_fields["token_delta"] != 0 + and measurement_fields["tuple_leak_window_count"] > 0 + and measurement_fields["intact_window_count"] == 0 + ): + labels.append("repeated_ngram_or_tuple_survival") + if ( + measurement_fields["token_delta"] != 0 + and measurement_fields["intact_window_count"] > 0 + and measurement_fields["tuple_leak_window_count"] > 0 + ): + labels.append("tokenizer_resynchronization_or_long_range_recovery") + if ( + measurement_fields["collapsed_equals_collapsed_source"] + and measurement_fields["token_delta"] != 0 + ): + labels.append("whitespace_only_channel") + if ( + measurement_fields["reachable_unselected_static_cover"] > 0 + and not measurement_fields["budget_exhausted"] + and not measurement_fields["closure_free"] + ): + labels.append("possible_scheduler_headroom") + if not labels and not measurement_fields["closure_free"]: + labels.append("residual_geometry") + return tuple(labels) + + +def measure_arm( + *, + arm_id: str, + source_sample_id: str, + source_text: str, + registry: TransformRegistry, + tokenizer: Any, + tokenizer_identity_hash: str, + ngram_len: int = 5, + budget: int = 14, +) -> Cycle7ArmMeasurement: + require_clean_string("arm_id", arm_id) + require_clean_string("source_sample_id", source_sample_id) + if not isinstance(source_text, str): + raise TypeError("source_text must be a string") + require_sha256("tokenizer_identity_hash", tokenizer_identity_hash) + require_int("ngram_len", ngram_len) + require_int("budget", budget) + enumeration = registry.enumerate(source_text) + scheduled = schedule_cover_greedy_v4( + source_sample_id=source_sample_id, + source_text=source_text, + registry=registry, + enumeration=enumeration, + tokenizer=tokenizer, + tokenizer_identity_hash=tokenizer_identity_hash, + ngram_len=ngram_len, + budget=budget, + ) + transformed = registry.apply(enumeration, scheduled.selected_candidate_ids) + collapsed = collapse_horizontal_ascii_whitespace(transformed.output_text) + collapsed_source = collapse_horizontal_ascii_whitespace(source_text) + source_tokens = _encode(tokenizer, source_text) + transformed_tokens = _encode(tokenizer, transformed.output_text) + collapsed_tokens = _encode(tokenizer, collapsed) + selected_rules = tuple( + candidate.rule_id + for candidate in enumeration.candidates + if candidate.candidate_id in set(scheduled.selected_candidate_ids) + ) + rejections = _rejection_counts(enumeration) + remaining_unselected = set(scheduled.unselected_candidate_ids) - set( + scheduled.conflict_excluded_candidate_ids + ) + candidate_exhausted = ( + scheduled.selected_candidate_count < budget + and not scheduled.closure_free + and not remaining_unselected + ) + repetition = PublicRepetitionGeometry.create(ngram_len=ngram_len, context_history_size=1024) + config = GeometryConfig.create( + tokenizer_identity_hash=tokenizer_identity_hash, + ngram_len=ngram_len, + repetition_mask_policy_id=repetition.policy_id, + ) + engine = CounterfactualGeometryEngine( + tokenizer=tokenizer, + config=config, + eligibility_policy=repetition.eligibility_policy, + ) + root = engine.build_root(source_sample_id=source_sample_id, source_text=source_text) + collapsed_eval = engine.evaluate_output( + root=root, + current_text=source_text, + output_text=collapsed, + candidate_id=sha256_json(("cycle7-collapse", source_sample_id, arm_id)), + rule_hash=registry.ruleset_hash, + visible_cost_class=0, + family="cycle7-collapse", + tier=0, + ) + collapsed_closure = compute_tuple_closure( + root=root.observations, + transformed_tokens=collapsed_tokens, + expected_output_token_hash=collapsed_eval.output_token_hash, + ) + reachable = 0 + if remaining_unselected and not scheduled.closure_free: + reachable = len(remaining_unselected) + fields = { + "candidate_count": scheduled.candidate_count, + "selected_count": scheduled.selected_candidate_count, + "budget_exhausted": scheduled.budget_exhausted, + "candidate_exhausted": candidate_exhausted, + "protected_blocked_count": rejections[CandidateRejectionReason.PROTECTED_OVERLAP.value], + "quote_blocked_count": rejections[CandidateRejectionReason.QUOTE_POLICY_BLOCKED.value], + "token_delta": len(transformed_tokens) - len(source_tokens), + "tuple_leak_window_count": scheduled.tuple_leak_window_count, + "intact_window_count": scheduled.intact_window_count, + "closure_free": scheduled.closure_free, + "collapsed_equals_collapsed_source": collapsed == collapsed_source, + "reachable_unselected_static_cover": reachable, + } + measurement = Cycle7ArmMeasurement( + algorithm_version=CYCLE7_INSTRUMENTATION_VERSION, + arm_id=arm_id, + source_sample_id=source_sample_id, + source_text_hash=sha256_text(source_text), + candidate_count=scheduled.candidate_count, + selected_count=scheduled.selected_candidate_count, + selected_rule_ids=selected_rules, + budget=budget, + budget_exhausted=scheduled.budget_exhausted, + candidate_exhausted=candidate_exhausted, + rejection_counts=rejections, + protected_blocked_count=fields["protected_blocked_count"], + quote_blocked_count=fields["quote_blocked_count"], + conflict_excluded_count=len(scheduled.conflict_excluded_candidate_ids), + unselected_count=len(scheduled.unselected_candidate_ids), + token_count_source=len(source_tokens), + token_count_transformed=len(transformed_tokens), + token_count_collapsed=len(collapsed_tokens), + token_delta=fields["token_delta"], + collapsed_token_delta=len(collapsed_tokens) - len(source_tokens), + changed_token_positions=_changed_positions(source_tokens, transformed_tokens), + root_window_count=scheduled.root_window_count, + intact_window_count=scheduled.intact_window_count, + tuple_leak_window_count=scheduled.tuple_leak_window_count, + closure_free=scheduled.closure_free, + collapsed_intact_window_count=collapsed_eval.surviving_count, + collapsed_tuple_leak_window_count=collapsed_closure.leaked_window_count, + collapsed_closure_free=collapsed_closure.closure_free, + collapsed_equals_collapsed_source=fields["collapsed_equals_collapsed_source"], + reachable_unselected_static_cover=reachable, + failure_classes=_failure_classes(fields), + transformed_text=transformed.output_text, + collapsed_text=collapsed, + measurement_hash="0" * 64, + ) + return replace(measurement, measurement_hash=sha256_json(measurement.payload())) diff --git a/fuckmark/cycle7/ledger.py b/fuckmark/cycle7/ledger.py new file mode 100644 index 00000000..401fc91a --- /dev/null +++ b/fuckmark/cycle7/ledger.py @@ -0,0 +1,110 @@ +from __future__ import annotations + +from collections.abc import Mapping, Sequence + +from .._validation import require_int +from ..hashing import sha256_json + + +CYCLE7_LEDGER_VERSION = "cycle7-seed-ledger-v1" +CYCLE7_EXPLORATORY_ROLE = "exploratory_development" +CYCLE7_VALIDATION_ROLE = "validation_development" +CYCLE7_CONFIRMATION_RESERVED_ROLE = "confirmation_reserved" + +# Frozen Cycle 6 confirmation rows. Spent. Not development data. +SPENT_CONFIRMATION_SEED_BASES = (760_000, 770_000, 780_000) +# Earlier Cycle 6 development 0/16 work. Spent. Not Cycle 7 development data. +SPENT_DEVELOPMENT_SEED_BASES = (720_000, 730_000) +# Other bound profile schedule seeds that must not be reused as Cycle 7 corpora. +BLOCKED_PROFILE_SEED_BASES = ( + 1_120_000, + 1_130_000, + 1_140_000, + 1_150_000, + 1_160_000, +) +# Historic TinyDev / calibration generators. Not Cycle 7 corpora. +BLOCKED_HISTORIC_SEED_BASES = (401_000, 402_000, 500_000, 61000) + +# Chosen as a fresh 10k block after Cycle 6 confirmation (780000) with a gap. +# Selected before any Cycle 7 detector look. Do not reassign after scoring. +CYCLE7_EXPLORATORY_SEED_BASE = 810_000 +CYCLE7_VALIDATION_SEED_BASE = 820_000 +CYCLE7_CONFIRMATION_RESERVED_SEED_BASES = (830_000, 840_000, 850_000) + +_ALL_BLOCKED = frozenset( + ( + *SPENT_CONFIRMATION_SEED_BASES, + *SPENT_DEVELOPMENT_SEED_BASES, + *BLOCKED_PROFILE_SEED_BASES, + *BLOCKED_HISTORIC_SEED_BASES, + ) +) + + +def cycle7_seed_ledger_payload() -> dict[str, object]: + return { + "algorithm_version": CYCLE7_LEDGER_VERSION, + "spent_confirmation_seed_bases": list(SPENT_CONFIRMATION_SEED_BASES), + "spent_development_seed_bases": list(SPENT_DEVELOPMENT_SEED_BASES), + "blocked_profile_seed_bases": list(BLOCKED_PROFILE_SEED_BASES), + "blocked_historic_seed_bases": list(BLOCKED_HISTORIC_SEED_BASES), + "exploratory_development_seed_base": CYCLE7_EXPLORATORY_SEED_BASE, + "validation_development_seed_base": CYCLE7_VALIDATION_SEED_BASE, + "confirmation_reserved_seed_bases": list(CYCLE7_CONFIRMATION_RESERVED_SEED_BASES), + "selection_rule": ( + "Confirmation-reserved seeds were chosen as unused 10k blocks " + "without inspecting detector outcomes. Do not promote exploratory " + "or validation seeds into confirmation after seeing scores." + ), + "spent_corpus_rule": ( + "Do not use 720000, 730000, 760000, 770000, or 780000 as " + "Cycle 7 development, tuning, or confirmation data." + ), + } + + +def cycle7_seed_ledger_hash() -> str: + return sha256_json(cycle7_seed_ledger_payload()) + + +def role_for_seed_base(seed_base: int) -> str | None: + require_int("seed_base", seed_base) + if seed_base == CYCLE7_EXPLORATORY_SEED_BASE: + return CYCLE7_EXPLORATORY_ROLE + if seed_base == CYCLE7_VALIDATION_SEED_BASE: + return CYCLE7_VALIDATION_ROLE + if seed_base in CYCLE7_CONFIRMATION_RESERVED_SEED_BASES: + return CYCLE7_CONFIRMATION_RESERVED_ROLE + return None + + +def assert_development_seed(seed_base: int, *, role: str) -> None: + require_int("seed_base", seed_base) + if not isinstance(role, str) or not role: + raise ValueError("role must be a non-empty string") + if seed_base in _ALL_BLOCKED: + raise ValueError("seed_base is spent or otherwise blocked for Cycle 7 development") + expected = role_for_seed_base(seed_base) + if expected is None: + raise ValueError("seed_base is not in the Cycle 7 ledger") + if expected == CYCLE7_CONFIRMATION_RESERVED_ROLE or role == CYCLE7_CONFIRMATION_RESERVED_ROLE: + raise ValueError("confirmation-reserved seeds must not be used for development") + if role == CYCLE7_EXPLORATORY_ROLE and expected != CYCLE7_EXPLORATORY_ROLE: + raise ValueError("seed_base is not the Cycle 7 exploratory development seed") + if role == CYCLE7_VALIDATION_ROLE and expected != CYCLE7_VALIDATION_ROLE: + raise ValueError("seed_base is not the Cycle 7 validation development seed") + + +def assert_seed_not_spent(seed_base: int) -> None: + require_int("seed_base", seed_base) + if seed_base in _ALL_BLOCKED: + raise ValueError("seed_base is spent or otherwise blocked for Cycle 7") + + +def ledger_roles() -> Mapping[str, Sequence[int]]: + return { + CYCLE7_EXPLORATORY_ROLE: (CYCLE7_EXPLORATORY_SEED_BASE,), + CYCLE7_VALIDATION_ROLE: (CYCLE7_VALIDATION_SEED_BASE,), + CYCLE7_CONFIRMATION_RESERVED_ROLE: CYCLE7_CONFIRMATION_RESERVED_SEED_BASES, + } diff --git a/fuckmark/cycle7/registry.py b/fuckmark/cycle7/registry.py new file mode 100644 index 00000000..f4a918d0 --- /dev/null +++ b/fuckmark/cycle7/registry.py @@ -0,0 +1,31 @@ +from __future__ import annotations + +from collections.abc import Sequence + +from ..transforms.cycle7_quote_policy import ( + CYCLE7_QUOTE_SAFE_DURABLE_POLICY_ID, + CYCLE7_QUOTE_SAFE_MIXED_POLICY_ID, +) +from ..transforms.effectiveness_profile import content_region_destruction_surface_rules +from ..transforms.registry import TransformRegistry +from .durable_rules import cycle7_durable_rules + + +CYCLE7_DURABLE_REGISTRY_ID = "cycle7-durable-contraction-orthography-v1" +CYCLE7_COMBINED_REGISTRY_ID = "cycle7-durable-plus-cycle6-spacing-v1" + + +def cycle7_durable_transform_registry(identifiers: Sequence[str] = ()) -> TransformRegistry: + return TransformRegistry( + cycle7_durable_rules(), + identifiers, + quote_policy_id=CYCLE7_QUOTE_SAFE_DURABLE_POLICY_ID, + ) + + +def cycle7_combined_transform_registry(identifiers: Sequence[str] = ()) -> TransformRegistry: + return TransformRegistry( + (*cycle7_durable_rules(), *content_region_destruction_surface_rules()), + identifiers, + quote_policy_id=CYCLE7_QUOTE_SAFE_MIXED_POLICY_ID, + ) diff --git a/fuckmark/cycle7/whitespace_collapse.py b/fuckmark/cycle7/whitespace_collapse.py new file mode 100644 index 00000000..43a4b3fc --- /dev/null +++ b/fuckmark/cycle7/whitespace_collapse.py @@ -0,0 +1,87 @@ +from __future__ import annotations + +from .._validation import require_clean_string +from ..hashing import sha256_text +from ..sanitizer_robustness import nfkc_normalize, strip_unicode_format_characters + + +WHITESPACE_COLLAPSE_VERSION = "whitespace-collapse-v1" +WHITESPACE_COLLAPSE_SPEC_ID = "ascii-horizontal-run-collapse-preserve-newlines-v1" + +CYCLE7_SANITIZER_VARIANT_IDS = ( + "raw", + "nfkc", + "cf_strip", + "nfkc_cf_strip", + "ws_collapse", + "ws_collapse_nfkc_cf_strip", +) + +_HORIZONTAL = frozenset((" ", "\t")) + + +def normalize_newlines(text: str) -> str: + if not isinstance(text, str): + raise TypeError("text must be a string") + return text.replace("\r\n", "\n").replace("\r", "\n") + + +def collapse_horizontal_ascii_whitespace(text: str) -> str: + """Collapse runs of U+0020 and U+0009 to a single U+0020 per line. + + Specification (whitespace-collapse-v1): + - Convert CR LF and CR to LF first. + - Preserve LF as line structure. Do not join lines. + - On each line, replace every maximal run of SPACE (U+0020) and HT (U+0009) + with exactly one SPACE. + - Do not trim leftover single leading or trailing spaces. + - Do not collapse other Unicode whitespace (including NBSP U+00A0). + - Do not collapse vertical whitespace other than the CR/LF normalization above. + """ + if not isinstance(text, str): + raise TypeError("text must be a string") + lines: list[str] = [] + for line in normalize_newlines(text).split("\n"): + pieces: list[str] = [] + index = 0 + length = len(line) + while index < length: + character = line[index] + if character in _HORIZONTAL: + pieces.append(" ") + index += 1 + while index < length and line[index] in _HORIZONTAL: + index += 1 + continue + pieces.append(character) + index += 1 + lines.append("".join(pieces)) + return "\n".join(lines) + + +def sanitize_cycle7_variant(variant_id: str, text: str) -> str: + require_clean_string("variant_id", variant_id) + if not isinstance(text, str): + raise TypeError("text must be a string") + if variant_id == "raw": + return text + if variant_id == "nfkc": + return nfkc_normalize(text) + if variant_id == "cf_strip": + return strip_unicode_format_characters(text) + if variant_id == "nfkc_cf_strip": + return strip_unicode_format_characters(nfkc_normalize(text)) + if variant_id == "ws_collapse": + return collapse_horizontal_ascii_whitespace(text) + if variant_id == "ws_collapse_nfkc_cf_strip": + collapsed = collapse_horizontal_ascii_whitespace(text) + normalized = strip_unicode_format_characters(nfkc_normalize(collapsed)) + return collapse_horizontal_ascii_whitespace(normalized) + raise ValueError(f"unknown Cycle 7 sanitizer variant id: {variant_id}") + + +def whitespace_collapse_identity_hash() -> str: + return sha256_text( + f"{WHITESPACE_COLLAPSE_VERSION}:{WHITESPACE_COLLAPSE_SPEC_ID}:" + f"{','.join(CYCLE7_SANITIZER_VARIANT_IDS)}" + ) diff --git a/fuckmark/cycle7_stage_a_hf.py b/fuckmark/cycle7_stage_a_hf.py new file mode 100644 index 00000000..cec11678 --- /dev/null +++ b/fuckmark/cycle7_stage_a_hf.py @@ -0,0 +1,391 @@ +from __future__ import annotations + +import argparse +import json +from pathlib import Path +from typing import Any + +from .adapters import HuggingFaceSynthIDAdapter, HuggingFaceSynthIDConfig +from .config import canonical_json_text +from .corpus.schema import CorpusDomain, WatermarkLabel +from .corpus.tiny_dev import TINY_DEV_TARGET_LENGTH +from .corpus.tiny_dev_generation import TINY_DEV_DEFAULT_MAX_ATTEMPTS, TINY_DEV_PAIR_SEED_STRIDE +from .cycle7.compare import ( + CYCLE6_SPACING_ARM_ID, + CYCLE7_COMBINED_ARM_ID, + CYCLE7_DURABLE_ARM_ID, + compare_arms_on_text, + run_fixture_stage_a, + summarize_arm, +) +from .cycle7.decision import classify_fixture_stage_a +from .cycle7.ledger import ( + CYCLE7_EXPLORATORY_ROLE, + CYCLE7_EXPLORATORY_SEED_BASE, + assert_development_seed, +) +from .cycle7.whitespace_collapse import CYCLE7_SANITIZER_VARIANT_IDS, sanitize_cycle7_variant +from .detectors import weighted_mean_evidence +from .durable_io import write_canonical_json_fsynced +from .experiments.cycle6_confirmation import CYCLE6_BUDGET, CYCLE6_THRESHOLD +from .hashing import sha256_json, sha256_text +from .native_observations import build_native_observations +from .tiny_dev_context_survival_plan_hf import DEFAULT_MODEL_ID, DEFAULT_MODEL_REVISION +from .tiny_dev_corpus_hf import HuggingFaceTinyDevBackend +from .tiny_dev_detector_hf import default_watermark_payload +from .tiny_dev_transform_hf import _encode_text + + +CYCLE7_STAGE_A_DETECTOR_VERSION = "cycle7-stage-a-detector-compare-v1" + +_STAGE_A_TEMPLATES = { + CorpusDomain.GENERAL_EXPLANATORY: ( + "Explain in plain English why {topic} matters in careful scientific work. " + "Write one coherent paragraph without a list." + ), + CorpusDomain.TECHNICAL_EXPLANATION: ( + "Give a technical explanation of {topic} in software experiments. " + "Define the main idea, one failure mode, and one validation check." + ), + CorpusDomain.CONVERSATIONAL_PROSE: ( + "Answer a colleague who asks why {topic} matters. " + "Use natural conversational prose while keeping the explanation precise." + ), + CorpusDomain.STRUCTURED_INSTRUCTIONAL: ( + "Write a short three-step instruction for applying {topic} in an experiment. " + "Use complete sentences and keep each step concrete." + ), +} + +_STAGE_A_TOPIC = "reproducibility" + + +class _OffsetTokenizer: + def encode(self, text, add_special_tokens=False): + return list(text.encode("utf-8")) + + def __call__(self, text, add_special_tokens=False, return_offsets_mapping=False): + data = text.encode("utf-8") + result = {"input_ids": list(data)} + if return_offsets_mapping: + result["offset_mapping"] = [(index, index + 1) for index in range(len(data))] + return result + + +def _score_text(sample_id: str, text: str, tokenizer: Any, adapter: HuggingFaceSynthIDAdapter, eos_token_id: int) -> float: + tokens = _encode_text(tokenizer, text) + batch = build_native_observations(sample_id, tokens, eos_token_id, adapter) + return weighted_mean_evidence(batch).raw_score + + +def _generate_stage_a_samples(backend: HuggingFaceTinyDevBackend, seed_base: int) -> tuple[dict[str, object], ...]: + assert_development_seed(seed_base, role=CYCLE7_EXPLORATORY_ROLE) + samples: list[dict[str, object]] = [] + for pair_index, domain in enumerate(CorpusDomain): + prompt = _STAGE_A_TEMPLATES[domain].format(topic=_STAGE_A_TOPIC) + pair_seed_base = seed_base + pair_index * TINY_DEV_PAIR_SEED_STRIDE + accepted = None + for attempt in range(TINY_DEV_DEFAULT_MAX_ATTEMPTS): + seed = pair_seed_base + attempt + control = backend.generate(prompt, seed, watermarked=False) + watermarked = backend.generate(prompt, seed, watermarked=True) + if len(control.continuation_token_ids) != TINY_DEV_TARGET_LENGTH: + continue + if len(watermarked.continuation_token_ids) != TINY_DEV_TARGET_LENGTH: + continue + if control.text == watermarked.text: + continue + accepted = (seed, control, watermarked) + break + if accepted is None: + raise RuntimeError(f"Cycle 7 Stage A failed to generate a pair for {domain.value}") + seed, control, watermarked = accepted + for label, generated in ( + (WatermarkLabel.UNWATERMARKED, control), + (WatermarkLabel.WATERMARKED, watermarked), + ): + sample_id = f"cycle7-stage-a-{domain.value}-{label.value}" + samples.append( + { + "sample_id": sample_id, + "domain": domain.value, + "label": label.value, + "prompt": prompt, + "seed": seed, + "text": generated.text, + "text_sha256": sha256_text(generated.text), + "text_only_token_ids": generated.text_only_token_ids, + } + ) + return tuple(samples) + + +def _detector_arm_summary(rows: tuple[dict[str, object], ...], arm_id: str) -> dict[str, object]: + watermarked = tuple(row for row in rows if row["label"] == WatermarkLabel.WATERMARKED.value and row["arm_id"] == arm_id) + unwatermarked = tuple(row for row in rows if row["label"] == WatermarkLabel.UNWATERMARKED.value and row["arm_id"] == arm_id) + def _count(selected, variant, field): + return sum(bool(row["sanitizers"][variant][field]) for row in selected) + def _mean(selected, variant): + return sum(float(row["sanitizers"][variant]["score"]) for row in selected) / len(selected) + payload = { + "arm_id": arm_id, + "watermarked_row_count": len(watermarked), + "unwatermarked_row_count": len(unwatermarked), + "pristine_watermarked_detected": sum(bool(row["pristine_detected"]) for row in watermarked), + "pristine_unwatermarked_detected": sum(bool(row["pristine_detected"]) for row in unwatermarked), + "raw_watermarked_detected": _count(watermarked, "raw", "detected"), + "raw_unwatermarked_detected": _count(unwatermarked, "raw", "detected"), + "ws_collapse_watermarked_detected": _count(watermarked, "ws_collapse", "detected"), + "ws_collapse_unwatermarked_detected": _count(unwatermarked, "ws_collapse", "detected"), + "raw_watermarked_mean_score": _mean(watermarked, "raw"), + "ws_collapse_watermarked_mean_score": _mean(watermarked, "ws_collapse"), + "median_raw_watermarked_score": sorted(float(row["sanitizers"]["raw"]["score"]) for row in watermarked)[len(watermarked) // 2], + "worst_raw_watermarked_score": max(float(row["sanitizers"]["raw"]["score"]) for row in watermarked), + } + return payload + + +def _evaluate_samples( + samples: tuple[dict[str, object], ...], + tokenizer: Any, + identity_hash: str, + adapter: HuggingFaceSynthIDAdapter, + eos: int, +) -> tuple[tuple[dict[str, object], ...], tuple[dict[str, object], ...], dict[str, object]]: + scored_rows: list[dict[str, object]] = [] + geometry_rows: list[dict[str, object]] = [] + for sample in samples: + arms = compare_arms_on_text( + source_sample_id=str(sample["sample_id"]), + source_text=str(sample["text"]), + tokenizer=tokenizer, + tokenizer_identity_hash=identity_hash, + budget=CYCLE6_BUDGET, + ) + geometry_rows.append( + { + "sample_id": sample["sample_id"], + "label": sample["label"], + "domain": sample["domain"], + "arms": {arm_id: summarize_arm(measurement) for arm_id, measurement in arms.items()}, + } + ) + pristine = _score_text( + f"{sample['sample_id']}-pristine", + str(sample["text"]), + tokenizer, + adapter, + eos, + ) + for arm_id, measurement in arms.items(): + sanitizers = {} + for variant in CYCLE7_SANITIZER_VARIANT_IDS: + text = sanitize_cycle7_variant(variant, measurement.transformed_text) + score = _score_text( + f"{sample['sample_id']}-{arm_id}-{variant}", + text, + tokenizer, + adapter, + eos, + ) + sanitizers[variant] = { + "text_hash": sha256_text(text), + "score": score, + "detected": score >= CYCLE6_THRESHOLD, + } + scored_rows.append( + { + "sample_id": sample["sample_id"], + "label": sample["label"], + "domain": sample["domain"], + "arm_id": arm_id, + "pristine_score": pristine, + "pristine_detected": pristine >= CYCLE6_THRESHOLD, + "geometry": summarize_arm(measurement), + "sanitizers": sanitizers, + } + ) + summaries = { + arm_id: _detector_arm_summary(tuple(scored_rows), arm_id) + for arm_id in (CYCLE6_SPACING_ARM_ID, CYCLE7_DURABLE_ARM_ID, CYCLE7_COMBINED_ARM_ID) + } + return tuple(geometry_rows), tuple(scored_rows), summaries + + +def _adapter_and_tokenizer(device: str) -> tuple[Any, Any, HuggingFaceSynthIDAdapter, str, int]: + backend = HuggingFaceTinyDevBackend( + DEFAULT_MODEL_ID, + DEFAULT_MODEL_REVISION, + device=device, + temperature=0.8, + top_k=50, + top_p=0.95, + ) + tokenizer = backend._tokenizer + tokenizer.padding_side = "left" + payload = default_watermark_payload() + adapter = HuggingFaceSynthIDAdapter.from_torch( + HuggingFaceSynthIDConfig( + ngram_len=int(payload["ngram_len"]), + keys=tuple(payload["keys"]), + context_history_size=int(payload["context_history_size"]), + sampling_table_seed=int(payload["sampling_table_seed"]), + sampling_table_size=int(payload["sampling_table_size"]), + skip_first_ngram_calls=bool(payload["skip_first_ngram_calls"]), + debug_mode=bool(payload["debug_mode"]), + ) + ) + eos = tokenizer.eos_token_id + if eos is None: + raise RuntimeError("GPT-2 tokenizer must define eos_token_id") + return backend, tokenizer, adapter, backend.model_identity.identity_hash, int(eos) + + +def _build_detector_artifact( + samples: tuple[dict[str, object], ...], + geometry_rows: tuple[dict[str, object], ...], + scored_rows: tuple[dict[str, object], ...], + summaries: dict[str, object], + *, + catalog_version: str, +) -> dict[str, object]: + artifact = { + "algorithm_version": CYCLE7_STAGE_A_DETECTOR_VERSION, + "seed_base": CYCLE7_EXPLORATORY_SEED_BASE, + "durable_catalog_version": catalog_version, + "model": DEFAULT_MODEL_ID, + "model_revision": DEFAULT_MODEL_REVISION, + "threshold": CYCLE6_THRESHOLD, + "budget": CYCLE6_BUDGET, + "sanitizer_ids": CYCLE7_SANITIZER_VARIANT_IDS, + "samples": tuple( + { + "sample_id": sample["sample_id"], + "domain": sample["domain"], + "label": sample["label"], + "seed": sample["seed"], + "text": sample["text"], + "text_sha256": sample["text_sha256"], + } + for sample in samples + ), + "geometry_rows": geometry_rows, + "scored_rows": scored_rows, + "summaries": summaries, + "detector_access_used_for_selection": False, + "secret_access_used_for_selection": False, + } + return {**artifact, "artifact_hash": sha256_json({k: v for k, v in artifact.items() if k != "artifact_hash"})} + + +def run_stage_a_detector_compare(*, device: str = "cpu") -> dict[str, object]: + from .cycle7.durable_rules import CYCLE7_DURABLE_RULE_CATALOG_VERSION + + assert_development_seed(CYCLE7_EXPLORATORY_SEED_BASE, role=CYCLE7_EXPLORATORY_ROLE) + backend, tokenizer, adapter, identity_hash, eos = _adapter_and_tokenizer(device) + samples = _generate_stage_a_samples(backend, CYCLE7_EXPLORATORY_SEED_BASE) + geometry_rows, scored_rows, summaries = _evaluate_samples( + samples, tokenizer, identity_hash, adapter, eos + ) + return _build_detector_artifact( + samples, + geometry_rows, + scored_rows, + summaries, + catalog_version=CYCLE7_DURABLE_RULE_CATALOG_VERSION, + ) + + +def rescore_stage_a_detector_compare(previous: dict[str, object], *, device: str = "cpu") -> dict[str, object]: + from .cycle7.durable_rules import CYCLE7_DURABLE_RULE_CATALOG_VERSION + from .tiny_dev_context_survival_plan_hf import runtime_tokenizer_identity_public + from transformers import AutoTokenizer + + assert_development_seed(CYCLE7_EXPLORATORY_SEED_BASE, role=CYCLE7_EXPLORATORY_ROLE) + samples = tuple(previous["samples"]) + tokenizer = AutoTokenizer.from_pretrained( + DEFAULT_MODEL_ID, + revision=DEFAULT_MODEL_REVISION, + padding_side="left", + ) + if tokenizer.pad_token_id is None: + tokenizer.pad_token = tokenizer.eos_token + identity = runtime_tokenizer_identity_public(tokenizer, DEFAULT_MODEL_ID, DEFAULT_MODEL_REVISION) + payload = default_watermark_payload() + adapter = HuggingFaceSynthIDAdapter.from_torch( + HuggingFaceSynthIDConfig( + ngram_len=int(payload["ngram_len"]), + keys=tuple(payload["keys"]), + context_history_size=int(payload["context_history_size"]), + sampling_table_seed=int(payload["sampling_table_seed"]), + sampling_table_size=int(payload["sampling_table_size"]), + skip_first_ngram_calls=bool(payload["skip_first_ngram_calls"]), + debug_mode=bool(payload["debug_mode"]), + ) + ) + eos = tokenizer.eos_token_id + if eos is None: + raise RuntimeError("GPT-2 tokenizer must define eos_token_id") + geometry_rows, scored_rows, summaries = _evaluate_samples( + samples, tokenizer, identity.identity_hash, adapter, int(eos) + ) + return _build_detector_artifact( + samples, + geometry_rows, + scored_rows, + summaries, + catalog_version=CYCLE7_DURABLE_RULE_CATALOG_VERSION, + ) + + +def main(argv: list[str] | None = None) -> int: + parser = argparse.ArgumentParser(prog="fuckmark-cycle7-stage-a") + parser.add_argument("--device", choices=("cpu", "cuda"), default="cpu") + parser.add_argument( + "--fixture-json", + type=Path, + default=Path("evidence/cycle7-stage-a-2026-08-25/fixture-geometry.json"), + ) + parser.add_argument( + "--detector-json", + type=Path, + default=Path("evidence/cycle7-stage-a-2026-08-25/detector-compare.json"), + ) + parser.add_argument("--skip-detector", action="store_true") + parser.add_argument( + "--rescore-from", + type=Path, + default=None, + help="Reuse previously generated Stage A texts; do not regenerate.", + ) + args = parser.parse_args(argv) + fixture = run_fixture_stage_a(_OffsetTokenizer()) + write_canonical_json_fsynced(args.fixture_json, fixture) + detector = None + if not args.skip_detector: + if args.rescore_from is not None: + previous = json.loads(args.rescore_from.read_text(encoding="utf-8")) + detector = rescore_stage_a_detector_compare(previous, device=args.device) + else: + detector = run_stage_a_detector_compare(device=args.device) + write_canonical_json_fsynced(args.detector_json, detector) + fixture = { + **{k: v for k, v in fixture.items() if k != "artifact_hash"}, + "detector": { + "available": True, + "cycle6_spacing": detector["summaries"][CYCLE6_SPACING_ARM_ID], + "durable": detector["summaries"][CYCLE7_DURABLE_ARM_ID], + "combined": detector["summaries"][CYCLE7_COMBINED_ARM_ID], + }, + } + fixture = {**fixture, "artifact_hash": sha256_json({k: v for k, v in fixture.items() if k != "artifact_hash"})} + write_canonical_json_fsynced(args.fixture_json, fixture) + decision = classify_fixture_stage_a(fixture) + args.fixture_json.parent.mkdir(parents=True, exist_ok=True) + write_canonical_json_fsynced(args.fixture_json.parent / "decision.json", decision) + print(canonical_json_text(decision)) + return 0 + + +if __name__ == "__main__": + raise SystemExit(main()) diff --git a/fuckmark/transforms/cycle7_quote_policy.py b/fuckmark/transforms/cycle7_quote_policy.py new file mode 100644 index 00000000..d993373c --- /dev/null +++ b/fuckmark/transforms/cycle7_quote_policy.py @@ -0,0 +1,130 @@ +from __future__ import annotations + +from collections.abc import Sequence +from typing import TYPE_CHECKING + +from .lexical_rules import LexicalTemplateRule +from .quote_policy import ( + QUOTE_SAFE_SURFACE_POLICY_ID, + is_quote_safe_surface_rule, + is_surface_editable_quotation, + quotation_spans, +) +from .rules import ( + GeneralWordLeadingSpacingRule, + GeneralWordSpacingRule, + LiteralTransformRule, + SurfaceSpacingRule, +) +from .schema import TransformFamily +from .syntax_rules import SyntaxTemplateRule + +if TYPE_CHECKING: + from .trace import TransformOperation + + +CYCLE7_QUOTE_SAFE_DURABLE_POLICY_ID = "quote-container-durable-v1" +CYCLE7_QUOTE_SAFE_MIXED_POLICY_ID = "quote-container-durable-or-spacing-v1" + +QUOTE_INTERIOR_POLICY_IDS = frozenset( + { + QUOTE_SAFE_SURFACE_POLICY_ID, + CYCLE7_QUOTE_SAFE_DURABLE_POLICY_ID, + CYCLE7_QUOTE_SAFE_MIXED_POLICY_ID, + } +) + +_DURABLE_RULE_PREFIXES = ( + "contract-", + "expand-", + "cycle7-contract-", + "cycle7-expand-", + "cycle7-ortho-", + "lexical-", + "syntax-", +) + + +def is_cycle7_quote_durable_rule(rule: object) -> bool: + if isinstance(rule, (GeneralWordLeadingSpacingRule, GeneralWordSpacingRule, SurfaceSpacingRule)): + return False + if isinstance(rule, LiteralTransformRule): + if rule.family is TransformFamily.CONTRACTION: + return getattr(rule, "rule_id", "").startswith( + ("contract-", "expand-", "cycle7-contract-", "cycle7-expand-") + ) + return rule.family is TransformFamily.ORTHOGRAPHY and getattr(rule, "rule_id", "").startswith( + "cycle7-ortho-" + ) + return isinstance(rule, (LexicalTemplateRule, SyntaxTemplateRule)) + + +def is_cycle7_quote_durable_rule_id(rule_id: str) -> bool: + if not isinstance(rule_id, str) or not rule_id: + return False + return rule_id.startswith(_DURABLE_RULE_PREFIXES) + + +def quote_interior_rule_allowed(quote_policy_id: str, rule: object) -> bool: + if quote_policy_id == QUOTE_SAFE_SURFACE_POLICY_ID: + return is_quote_safe_surface_rule(rule) + if quote_policy_id == CYCLE7_QUOTE_SAFE_DURABLE_POLICY_ID: + return is_cycle7_quote_durable_rule(rule) + if quote_policy_id == CYCLE7_QUOTE_SAFE_MIXED_POLICY_ID: + return is_quote_safe_surface_rule(rule) or is_cycle7_quote_durable_rule(rule) + return False + + +def _quote_overlaps_for_operation( + quotes: Sequence[object], + operation: TransformOperation, +) -> tuple[object, ...]: + return tuple( + span + for span in quotes + if operation.source_start < span.end and span.start < operation.source_end # type: ignore[attr-defined] + ) + + +def validate_cycle7_quote_operations( + source_text: str, + operations: Sequence[TransformOperation], + quote_policy_id: str, +) -> None: + from .protected import ProtectedSpanExtractor + + if quote_policy_id not in ( + CYCLE7_QUOTE_SAFE_DURABLE_POLICY_ID, + CYCLE7_QUOTE_SAFE_MIXED_POLICY_ID, + ): + raise ValueError("unsupported Cycle 7 quotation protection policy") + manifest = ProtectedSpanExtractor().extract(source_text) + quotes = quotation_spans(manifest.spans) + mixed = quote_policy_id == CYCLE7_QUOTE_SAFE_MIXED_POLICY_ID + for operation in operations: + overlaps = _quote_overlaps_for_operation(quotes, operation) + if not overlaps: + continue + if len(overlaps) != 1 or not is_surface_editable_quotation(overlaps[0]): + raise ValueError("quote-safe operation must stay inside one quotation container") + quote = overlaps[0] + if operation.source_start <= quote.start or operation.source_end >= quote.end: + raise ValueError("quote-safe operation must not alter quotation delimiters") + spacing = operation.rule_id.startswith("surface-space-") + durable = is_cycle7_quote_durable_rule_id(operation.rule_id) + if mixed and spacing: + if operation.after_text not in ( + operation.before_text + " ", + " " + operation.before_text, + ): + raise ValueError("quote-safe surface operation must add exactly one ASCII space") + continue + if not durable: + raise ValueError("only durable Cycle 7 rules may operate inside quotations under this policy") + if operation.after_text == operation.before_text: + raise ValueError("quote-safe durable operation must change the quoted interior") + if operation.after_text in ( + operation.before_text + " ", + " " + operation.before_text, + ) and operation.after_text.replace(" ", "") == operation.before_text.replace(" ", ""): + raise ValueError("quote-safe durable policy must not use spacing-only rewrites") diff --git a/fuckmark/transforms/registry.py b/fuckmark/transforms/registry.py index 2aae9c11..09df71eb 100644 --- a/fuckmark/transforms/registry.py +++ b/fuckmark/transforms/registry.py @@ -11,11 +11,14 @@ from .lexical_rules import LexicalTemplateRule, development_lexical_rules from .protected import ProtectedSpanExtractor from .protected_artifacts import ProtectedSpan, UserProtectedRange +from .cycle7_quote_policy import ( + QUOTE_INTERIOR_POLICY_IDS, + quote_interior_rule_allowed, +) from .quote_policy import ( BLANKET_QUOTE_PROTECTION_POLICY_ID, QUOTE_SAFE_SURFACE_POLICY_ID, is_surface_editable_quotation, - is_quote_safe_surface_rule, quotation_spans, ) from .rules import TransformRule, default_contraction_rules, validate_rules @@ -79,6 +82,7 @@ def __init__( if quote_policy_id not in ( BLANKET_QUOTE_PROTECTION_POLICY_ID, QUOTE_SAFE_SURFACE_POLICY_ID, + *sorted(QUOTE_INTERIOR_POLICY_IDS - {QUOTE_SAFE_SURFACE_POLICY_ID}), ): raise ValueError("unsupported quotation protection policy") self._quote_policy_id = quote_policy_id @@ -112,11 +116,11 @@ def enumerate(self, text: str, user_ranges: Sequence[UserProtectedRange] = ()) - raise TypeError("text must be a string") if len(self._rules) * len(text) > _MAX_RULE_SCAN_WORK: raise ValueError("transform rule scanning exceeded work limit") - quote_safe = self._quote_policy_id == QUOTE_SAFE_SURFACE_POLICY_ID + quote_interior = self._quote_policy_id in QUOTE_INTERIOR_POLICY_IDS protected = self._extractor.extract( text, user_ranges, - include_quotations=not quote_safe, + include_quotations=not quote_interior, ) quotation_manifest = self._extractor.extract(text, include_quotations=True) quotes = quotation_spans(quotation_manifest.spans) @@ -139,8 +143,7 @@ def enumerate(self, text: str, user_ranges: Sequence[UserProtectedRange] = ()) - span for span in quotes if start < span.end and span.start < end ) if quote_overlaps and not ( - quote_safe - and is_quote_safe_surface_rule(rule) + quote_interior_rule_allowed(self._quote_policy_id, rule) and len(quote_overlaps) == 1 and is_surface_editable_quotation(quote_overlaps[0]) and start > quote_overlaps[0].start @@ -234,13 +237,13 @@ def apply(self, enumeration: CandidateEnumeration, candidate_ids: Sequence[str], output_hash = sha256_text(output_text) if selected and output_hash == input_hash: raise ValueError("non-empty candidate selection produced no net text change") - quote_safe = self._quote_policy_id == QUOTE_SAFE_SURFACE_POLICY_ID + quote_interior = self._quote_policy_id in QUOTE_INTERIOR_POLICY_IDS invariant_report = validate_hard_invariants( enumeration.input_text, output_text, self.identifiers, enumeration.protected_manifest.user_ranges, - include_quotations=not quote_safe, + include_quotations=not quote_interior, ) if invariant_report.status is not InvariantStatus.PASS: raise ValueError("transformation violated hard content invariants") @@ -248,8 +251,8 @@ def apply(self, enumeration: CandidateEnumeration, candidate_ids: Sequence[str], operation_tuple = tuple(operations) selection_policy_id = ( "explicit-candidate-ids-v4" - if not quote_safe - else "explicit-candidate-ids-v4:quote-container-surface-spacing-v1" + if not quote_interior + else f"explicit-candidate-ids-v4:{self._quote_policy_id}" ) trace_payload = {"algorithm_version": TRANSFORM_APPLY_ALGORITHM_VERSION, "registry_version": TRANSFORM_REGISTRY_ALGORITHM_VERSION, "selection_policy_id": selection_policy_id, "seed": seed, "input_hash": input_hash, "output_hash": output_hash, "ruleset_hash": self._ruleset_hash, "enumeration_hash": enumeration.enumeration_hash, "selected_candidate_ids": selected_tuple, "operations": operation_tuple, "precondition_failures": enumeration.rejections, "protected_span_violation_count": 0, "invariant_report": invariant_report} trace = TransformationTrace(TRANSFORM_APPLY_ALGORITHM_VERSION, TRANSFORM_REGISTRY_ALGORITHM_VERSION, selection_policy_id, seed, input_hash, output_hash, self._ruleset_hash, enumeration.enumeration_hash, selected_tuple, operation_tuple, enumeration.rejections, 0, invariant_report, sha256_json(trace_payload)) diff --git a/fuckmark/transforms/trace.py b/fuckmark/transforms/trace.py index e837e12c..1024d241 100644 --- a/fuckmark/transforms/trace.py +++ b/fuckmark/transforms/trace.py @@ -7,7 +7,12 @@ from .candidate_artifacts import CandidateRejection from .hard_invariants import HardInvariantReport, hard_invariant_signature from .invariants import validate_protected_invariants -from .quote_policy import validate_quote_safe_surface_operations +from .cycle7_quote_policy import ( + CYCLE7_QUOTE_SAFE_DURABLE_POLICY_ID, + CYCLE7_QUOTE_SAFE_MIXED_POLICY_ID, + validate_cycle7_quote_operations, +) +from .quote_policy import QUOTE_SAFE_SURFACE_POLICY_ID, validate_quote_safe_surface_operations from .schema import InvariantStatus @dataclass(frozen=True, slots=True) @@ -188,15 +193,25 @@ def __post_init__(self) -> None: raise ValueError("trace original hard-invariant signature does not match reconstructed input") if hard_invariant_signature(self.output_text) != self.trace.invariant_report.transformed_signature: raise ValueError("trace transformed hard-invariant signature does not match output_text") - quote_safe = self.trace.selection_policy_id.endswith( - ":quote-container-surface-spacing-v1" + quote_spacing = self.trace.selection_policy_id.endswith( + f":{QUOTE_SAFE_SURFACE_POLICY_ID}" ) - if quote_safe: + quote_cycle7 = self.trace.selection_policy_id.endswith( + (f":{CYCLE7_QUOTE_SAFE_DURABLE_POLICY_ID}", f":{CYCLE7_QUOTE_SAFE_MIXED_POLICY_ID}") + ) + if quote_spacing: validate_quote_safe_surface_operations(reconstructed_input, self.trace.operations) + elif quote_cycle7: + policy_id = self.trace.selection_policy_id.split(":", 1)[1] + validate_cycle7_quote_operations( + reconstructed_input, + self.trace.operations, + policy_id, + ) if validate_protected_invariants( reconstructed_input, self.output_text, - include_quotations=not quote_safe, + include_quotations=not (quote_spacing or quote_cycle7), ).status is not InvariantStatus.PASS: raise ValueError("transform result violates default protected-content invariants") require_sha256("result_hash", self.result_hash) diff --git a/tests/test_cycle7_durable_rules.py b/tests/test_cycle7_durable_rules.py new file mode 100644 index 00000000..a8c83110 --- /dev/null +++ b/tests/test_cycle7_durable_rules.py @@ -0,0 +1,139 @@ +from fuckmark.cycle7.durable_rules import ( + cycle7_bounded_copula_rules, + cycle7_new_contraction_rules, + cycle7_orthography_rules, + rejected_ambiguous_contraction_examples, +) +from fuckmark.cycle7.registry import cycle7_durable_transform_registry +from fuckmark.cycle7.whitespace_collapse import collapse_horizontal_ascii_whitespace +from fuckmark.experiments.cycle6_confirmation import CYCLE6_RULESET_HASH +from fuckmark.transforms import ProtectedSpanExtractor, quote_safe_zrd_transform_registry +from fuckmark.transforms.schema import CandidateRejectionReason, TransformFamily + + +def test_cycle6_quote_safe_ruleset_hash_is_unchanged() -> None: + assert quote_safe_zrd_transform_registry().ruleset_hash == CYCLE6_RULESET_HASH + + +def test_cycle7_catalog_excludes_ambiguous_contractions() -> None: + catalog = (*cycle7_new_contraction_rules(), *cycle7_bounded_copula_rules()) + pairs = {(rule.source.casefold(), rule.replacement.casefold()) for rule in catalog} + for source, replacement in rejected_ambiguous_contraction_examples(): + assert (source.casefold(), replacement.casefold()) not in pairs + assert not any(rule.source.casefold() == "it's been" for rule in catalog) + + +def test_have_contractions_are_reversible_and_survive_collapse() -> None: + registry = cycle7_durable_transform_registry() + text = "I have results and they have time." + enumeration = registry.enumerate(text) + selected = tuple( + candidate.candidate_id + for candidate in enumeration.candidates + if candidate.rule_id in {"cycle7-contract-i-have", "cycle7-contract-they-have"} + ) + assert len(selected) == 2 + result = registry.apply(enumeration, selected) + assert "I've" in result.output_text + assert "they've" in result.output_text + assert collapse_horizontal_ascii_whitespace(result.output_text) == result.output_text + inverse = registry.enumerate(result.output_text) + reverse = tuple( + candidate.candidate_id + for candidate in inverse.candidates + if candidate.rule_id in {"cycle7-expand-i-have", "cycle7-expand-they-have"} + ) + restored = registry.apply(inverse, reverse) + assert restored.output_text == text + + +def test_bounded_its_important_is_unambiguous_and_ignores_its_been() -> None: + registry = cycle7_durable_transform_registry() + enumeration = registry.enumerate("It's important to keep logs and it's been archived.") + selected = tuple( + candidate + for candidate in enumeration.candidates + if candidate.rule_id == "cycle7-expand-its-important" + ) + assert len(selected) == 1 + assert selected[0].source_text == "It's important" + result = registry.apply(enumeration, (selected[0].candidate_id,)) + assert "It is important to keep logs" in result.output_text + assert "it's been archived" in result.output_text + assert not any( + candidate.source_text.casefold() == "it's been" for candidate in enumeration.candidates + ) + assert not any(candidate.rule_id.endswith("its-not") for candidate in enumeration.candidates) + registry = cycle7_durable_transform_registry() + enumeration = registry.enumerate("We do not stop.") + forward = next(candidate for candidate in enumeration.candidates if candidate.rule_id == "contract-do-not") + result = registry.apply(enumeration, (forward.candidate_id,)) + assert result.output_text == "We don't stop." + assert collapse_horizontal_ascii_whitespace(result.output_text) == "We don't stop." + + +def test_orthography_towards_among_is_reversible() -> None: + registry = cycle7_durable_transform_registry() + text = "Walk towards the gate amongst the trees." + enumeration = registry.enumerate(text) + selected = tuple( + candidate.candidate_id + for candidate in enumeration.candidates + if candidate.rule_id in {"cycle7-ortho-towards-toward", "cycle7-ortho-amongst-among"} + ) + assert len(selected) == 2 + result = registry.apply(enumeration, selected) + assert result.output_text == "Walk toward the gate among the trees." + ids = {rule.rule_id for rule in cycle7_orthography_rules()} + assert "cycle7-ortho-toward-towards" in ids + + +def test_false_positive_grammar_and_all_caps_are_blocked() -> None: + registry = cycle7_durable_transform_registry() + assert not any( + candidate.rule_id == "contract-do-not" + for candidate in registry.enumerate("The donor notes were archived.").candidates + ) + enumeration = registry.enumerate("WE DO NOT STOP.") + assert any( + rejection.rule_id == "contract-do-not" + and rejection.reason is CandidateRejectionReason.ALL_CAPS_BLOCKED + for rejection in enumeration.rejections + ) + + +def test_protected_spans_block_durable_rewrites() -> None: + registry = cycle7_durable_transform_registry(("CamelCaseThing",)) + cases = ( + "Visit https://example.com/path and continue.", + "The value 12 is exact.", + "On 2024-08-25 we do not change the timestamp.", + "Run `do not` as a literal command token.", + "Keep CamelCaseThing intact while they are waiting.", + "The binary /usr/bin/true stays put.", + r"File C:\Windows\System32\drivers stays put.", + "Digest deadbeefcafebabe0123456789abcdef0123456789abcdef0123456789abcdef is fixed.", + "Prior work [1] remains cited.", + "Write lab@example.com immediately.", + ) + for text in cases: + enumeration = registry.enumerate(text) + protected = ProtectedSpanExtractor().extract(text).spans + for candidate in enumeration.candidates: + assert all( + not (candidate.start < span.end and span.start < candidate.end) + for span in protected + ), (text, candidate.rule_id, candidate.source_text) + + +def test_punctuation_url_and_number_boundaries_do_not_fire_inside_tokens() -> None: + registry = cycle7_durable_transform_registry() + enumeration = registry.enumerate("See do-not-touch and donor.") + assert not any(candidate.rule_id == "contract-do-not" for candidate in enumeration.candidates) + + +def test_durable_rules_are_not_spacing_family() -> None: + for rule in (*cycle7_new_contraction_rules(), *cycle7_orthography_rules()): + assert rule.family in {TransformFamily.CONTRACTION, TransformFamily.ORTHOGRAPHY} + assert not rule.source.endswith(" ") + assert " " not in rule.replacement diff --git a/tests/test_cycle7_instrumentation.py b/tests/test_cycle7_instrumentation.py new file mode 100644 index 00000000..f504130c --- /dev/null +++ b/tests/test_cycle7_instrumentation.py @@ -0,0 +1,83 @@ +from fuckmark.cycle7.compare import ( + CYCLE6_SPACING_ARM_ID, + CYCLE7_DURABLE_ARM_ID, + run_fixture_stage_a, +) +from fuckmark.cycle7.decision import PROMISING_DEVELOPMENT, classify_fixture_stage_a +from fuckmark.cycle7.fixtures import CONTRACTION_RICH, CONTRACTION_SPARSE +from fuckmark.cycle7.instrumentation import measure_arm +from fuckmark.cycle7.registry import cycle7_durable_transform_registry +from fuckmark.cycle7.whitespace_collapse import collapse_horizontal_ascii_whitespace +from fuckmark.hashing import sha256_text +from fuckmark.transforms import quote_safe_zrd_transform_registry + + +class _OffsetTokenizer: + def encode(self, text, add_special_tokens=False): + return list(text.encode("utf-8")) + + def __call__(self, text, add_special_tokens=False, return_offsets_mapping=False): + data = text.encode("utf-8") + result = {"input_ids": list(data)} + if return_offsets_mapping: + result["offset_mapping"] = [(index, index + 1) for index in range(len(data))] + return result + + +TOKENIZER_HASH = sha256_text("cycle7-fixture-tokenizer") + + +def test_instrumentation_records_density_geometry_and_collapse() -> None: + measurement = measure_arm( + arm_id="cycle7_durable", + source_sample_id="rich", + source_text=CONTRACTION_RICH, + registry=cycle7_durable_transform_registry(), + tokenizer=_OffsetTokenizer(), + tokenizer_identity_hash=TOKENIZER_HASH, + budget=14, + ) + assert measurement.candidate_count > 0 + assert measurement.selected_count > 0 + assert measurement.measurement_hash == __import__( + "fuckmark.hashing", fromlist=["sha256_json"] + ).sha256_json(measurement.payload()) + assert measurement.collapsed_equals_collapsed_source is False + assert "insufficient_transform_density" not in measurement.failure_classes + assert measurement.root_window_count >= 0 + assert measurement.quote_blocked_count >= 0 + assert measurement.protected_blocked_count >= 0 + + +def test_cycle6_spacing_is_undone_by_whitespace_collapse_on_rich_text() -> None: + probe = "Careful testing matters before any claim becomes knowledge." + measurement = measure_arm( + arm_id="cycle6_spacing_b14", + source_sample_id="spacing-probe", + source_text=probe, + registry=quote_safe_zrd_transform_registry(), + tokenizer=_OffsetTokenizer(), + tokenizer_identity_hash=TOKENIZER_HASH, + budget=14, + ) + if measurement.selected_count and all( + rule_id.startswith("surface-space-") for rule_id in measurement.selected_rule_ids + ): + assert measurement.collapsed_equals_collapsed_source is True + assert collapse_horizontal_ascii_whitespace(measurement.transformed_text) == ( + collapse_horizontal_ascii_whitespace(probe) + ) + + +def test_fixture_stage_a_compares_three_arms_and_classifies() -> None: + report = run_fixture_stage_a(_OffsetTokenizer(), TOKENIZER_HASH) + assert report["artifact_hash"] + rich = next(row for row in report["rows"] if row["source_sample_id"] == "contraction-rich") + sparse = next(row for row in report["rows"] if row["source_sample_id"] == "contraction-sparse") + assert rich["arms"][CYCLE7_DURABLE_ARM_ID]["selected_count"] > 0 + assert rich["arms"][CYCLE7_DURABLE_ARM_ID]["collapsed_equals_collapsed_source"] is False + assert CYCLE6_SPACING_ARM_ID in rich["arms"] + assert sparse["source_sample_id"] == "contraction-sparse" + decision = classify_fixture_stage_a(report) + assert decision["decision"] == PROMISING_DEVELOPMENT + assert CONTRACTION_SPARSE diff --git a/tests/test_cycle7_ledger.py b/tests/test_cycle7_ledger.py new file mode 100644 index 00000000..1f60338d --- /dev/null +++ b/tests/test_cycle7_ledger.py @@ -0,0 +1,44 @@ +from fuckmark.cycle7.ledger import ( + CYCLE7_CONFIRMATION_RESERVED_SEED_BASES, + CYCLE7_EXPLORATORY_ROLE, + CYCLE7_EXPLORATORY_SEED_BASE, + CYCLE7_VALIDATION_SEED_BASE, + SPENT_CONFIRMATION_SEED_BASES, + SPENT_DEVELOPMENT_SEED_BASES, + assert_development_seed, + assert_seed_not_spent, + cycle7_seed_ledger_hash, + cycle7_seed_ledger_payload, +) +import pytest + + +def test_spent_and_reserved_seeds_are_disjoint_and_documented() -> None: + payload = cycle7_seed_ledger_payload() + assert tuple(payload["spent_confirmation_seed_bases"]) == SPENT_CONFIRMATION_SEED_BASES + assert SPENT_CONFIRMATION_SEED_BASES == (760000, 770000, 780000) + assert SPENT_DEVELOPMENT_SEED_BASES == (720000, 730000) + assert CYCLE7_EXPLORATORY_SEED_BASE == 810000 + assert CYCLE7_VALIDATION_SEED_BASE == 820000 + assert CYCLE7_CONFIRMATION_RESERVED_SEED_BASES == (830000, 840000, 850000) + blocked = { + *SPENT_CONFIRMATION_SEED_BASES, + *SPENT_DEVELOPMENT_SEED_BASES, + CYCLE7_EXPLORATORY_SEED_BASE, + CYCLE7_VALIDATION_SEED_BASE, + *CYCLE7_CONFIRMATION_RESERVED_SEED_BASES, + } + assert len(blocked) == 10 + assert cycle7_seed_ledger_hash() == cycle7_seed_ledger_hash() + + +def test_exploratory_seed_is_admitted_and_spent_seeds_are_rejected() -> None: + assert_development_seed(CYCLE7_EXPLORATORY_SEED_BASE, role=CYCLE7_EXPLORATORY_ROLE) + with pytest.raises(ValueError, match="spent"): + assert_development_seed(760000, role=CYCLE7_EXPLORATORY_ROLE) + with pytest.raises(ValueError, match="spent"): + assert_seed_not_spent(720000) + with pytest.raises(ValueError, match="confirmation-reserved"): + assert_development_seed(830000, role=CYCLE7_EXPLORATORY_ROLE) + with pytest.raises(ValueError, match="exploratory"): + assert_development_seed(CYCLE7_VALIDATION_SEED_BASE, role=CYCLE7_EXPLORATORY_ROLE) diff --git a/tests/test_cycle7_quote_policy.py b/tests/test_cycle7_quote_policy.py new file mode 100644 index 00000000..95d2d1aa --- /dev/null +++ b/tests/test_cycle7_quote_policy.py @@ -0,0 +1,60 @@ +from fuckmark.cycle7.registry import ( + cycle7_combined_transform_registry, + cycle7_durable_transform_registry, +) +from fuckmark.transforms import quote_safe_zrd_transform_registry +from fuckmark.transforms.cycle7_quote_policy import ( + CYCLE7_QUOTE_SAFE_DURABLE_POLICY_ID, + CYCLE7_QUOTE_SAFE_MIXED_POLICY_ID, +) +from fuckmark.transforms.schema import CandidateRejectionReason + + +QUOTE = 'He answered, "They are not finished and we do not agree."' + + +def test_cycle6_quote_safe_still_blocks_contractions_inside_quotes() -> None: + enumeration = quote_safe_zrd_transform_registry().enumerate(QUOTE) + assert any( + rejection.rule_id == "contract-do-not" + and rejection.reason is CandidateRejectionReason.QUOTE_POLICY_BLOCKED + for rejection in enumeration.rejections + ) + assert not any(candidate.rule_id == "contract-do-not" for candidate in enumeration.candidates) + assert any(candidate.rule_id.startswith("surface-space-") for candidate in enumeration.candidates) + + +def test_cycle7_durable_policy_allows_contractions_inside_quotes() -> None: + registry = cycle7_durable_transform_registry() + assert registry.quote_policy_id == CYCLE7_QUOTE_SAFE_DURABLE_POLICY_ID + enumeration = registry.enumerate(QUOTE) + selected = tuple( + candidate.candidate_id + for candidate in enumeration.candidates + if candidate.rule_id == "contract-do-not" + ) + assert selected + result = registry.apply(enumeration, selected) + assert result.output_text.startswith('He answered, "') + assert result.output_text.endswith('."') + assert "don't" in result.output_text or "aren't" in result.output_text + assert result.trace.selection_policy_id.endswith(f":{CYCLE7_QUOTE_SAFE_DURABLE_POLICY_ID}") + + +def test_cycle7_durable_policy_does_not_alter_quote_delimiters() -> None: + registry = cycle7_durable_transform_registry() + enumeration = registry.enumerate(QUOTE) + if enumeration.candidates: + result = registry.apply( + enumeration, + (enumeration.candidates[0].candidate_id,), + ) + assert result.output_text.count('"') == QUOTE.count('"') + + +def test_mixed_policy_allows_spacing_or_durable_inside_quotes() -> None: + registry = cycle7_combined_transform_registry() + assert registry.quote_policy_id == CYCLE7_QUOTE_SAFE_MIXED_POLICY_ID + enumeration = registry.enumerate(QUOTE) + assert any(candidate.rule_id.startswith("surface-space-") for candidate in enumeration.candidates) + assert any(candidate.rule_id == "contract-do-not" for candidate in enumeration.candidates) diff --git a/tests/test_cycle7_sanitizer_boundary.py b/tests/test_cycle7_sanitizer_boundary.py new file mode 100644 index 00000000..4654704c --- /dev/null +++ b/tests/test_cycle7_sanitizer_boundary.py @@ -0,0 +1,11 @@ +from fuckmark.cycle7.whitespace_collapse import CYCLE7_SANITIZER_VARIANT_IDS +from fuckmark.experiments.cycle6_confirmation import CYCLE6_SANITIZER_IDS +from fuckmark.sanitizer_robustness import SANITIZER_VARIANT_IDS + + +def test_cycle6_sanitizer_ids_are_untouched() -> None: + assert SANITIZER_VARIANT_IDS == ("raw", "nfkc", "cf_strip", "nfkc_cf_strip") + assert CYCLE6_SANITIZER_IDS == SANITIZER_VARIANT_IDS + assert CYCLE7_SANITIZER_VARIANT_IDS[:4] == SANITIZER_VARIANT_IDS + assert "ws_collapse" in CYCLE7_SANITIZER_VARIANT_IDS + assert "ws_collapse_nfkc_cf_strip" in CYCLE7_SANITIZER_VARIANT_IDS diff --git a/tests/test_cycle7_whitespace_collapse.py b/tests/test_cycle7_whitespace_collapse.py new file mode 100644 index 00000000..fc7f96ad --- /dev/null +++ b/tests/test_cycle7_whitespace_collapse.py @@ -0,0 +1,48 @@ +import unicodedata + +from fuckmark.cycle7.whitespace_collapse import ( + CYCLE7_SANITIZER_VARIANT_IDS, + WHITESPACE_COLLAPSE_VERSION, + collapse_horizontal_ascii_whitespace, + sanitize_cycle7_variant, +) +from fuckmark.sanitizer_robustness import nfkc_normalize, strip_unicode_format_characters + + +def test_whitespace_collapse_version_and_variant_ids_are_frozen() -> None: + assert WHITESPACE_COLLAPSE_VERSION == "whitespace-collapse-v1" + assert CYCLE7_SANITIZER_VARIANT_IDS == ( + "raw", + "nfkc", + "cf_strip", + "nfkc_cf_strip", + "ws_collapse", + "ws_collapse_nfkc_cf_strip", + ) + + +def test_collapse_repeated_ascii_spaces_and_tabs_but_preserve_newlines() -> None: + assert collapse_horizontal_ascii_whitespace("a b") == "a b" + assert collapse_horizontal_ascii_whitespace("a\t\tb") == "a b" + assert collapse_horizontal_ascii_whitespace("a \t b") == "a b" + assert collapse_horizontal_ascii_whitespace("a\nb c") == "a\nb c" + assert collapse_horizontal_ascii_whitespace("a\r\nb") == "a\nb" + assert collapse_horizontal_ascii_whitespace(" a ") == " a " + assert collapse_horizontal_ascii_whitespace("a\n\nb") == "a\n\nb" + + +def test_collapse_does_not_touch_nbsp_or_cycle6_nfkc_cf_arms() -> None: + nbsp = "a\u00a0\u00a0b" + assert collapse_horizontal_ascii_whitespace(nbsp) == nbsp + cf = "a\u200bb" + assert sanitize_cycle7_variant("cf_strip", cf) == strip_unicode_format_characters(cf) + assert sanitize_cycle7_variant("nfkc", "fi") == nfkc_normalize("fi") + assert sanitize_cycle7_variant("raw", "a b") == "a b" + + +def test_combined_collapse_runs_after_nfkc() -> None: + text = "a b\u200c c" + combined = sanitize_cycle7_variant("ws_collapse_nfkc_cf_strip", text) + assert " " not in combined + assert "\u200c" not in combined + assert unicodedata.normalize("NFKC", combined) == combined From 6e587ea016cf5a2f4e7845a5550f97e0b5d92b34 Mon Sep 17 00:00:00 2001 From: byte271 Date: Tue, 25 Aug 2026 15:46:03 -0400 Subject: [PATCH 2/3] Strip Python comments so repository hygiene and Release Readiness can pass. Co-authored-by: Cursor --- fuckmark/cycle7/ledger.py | 6 ------ fuckmark/cycle7/whitespace_collapse.py | 11 ----------- fuckmark/transforms/cycle7_quote_policy.py | 8 ++++---- 3 files changed, 4 insertions(+), 21 deletions(-) diff --git a/fuckmark/cycle7/ledger.py b/fuckmark/cycle7/ledger.py index 401fc91a..d27f65a6 100644 --- a/fuckmark/cycle7/ledger.py +++ b/fuckmark/cycle7/ledger.py @@ -11,11 +11,8 @@ CYCLE7_VALIDATION_ROLE = "validation_development" CYCLE7_CONFIRMATION_RESERVED_ROLE = "confirmation_reserved" -# Frozen Cycle 6 confirmation rows. Spent. Not development data. SPENT_CONFIRMATION_SEED_BASES = (760_000, 770_000, 780_000) -# Earlier Cycle 6 development 0/16 work. Spent. Not Cycle 7 development data. SPENT_DEVELOPMENT_SEED_BASES = (720_000, 730_000) -# Other bound profile schedule seeds that must not be reused as Cycle 7 corpora. BLOCKED_PROFILE_SEED_BASES = ( 1_120_000, 1_130_000, @@ -23,11 +20,8 @@ 1_150_000, 1_160_000, ) -# Historic TinyDev / calibration generators. Not Cycle 7 corpora. BLOCKED_HISTORIC_SEED_BASES = (401_000, 402_000, 500_000, 61000) -# Chosen as a fresh 10k block after Cycle 6 confirmation (780000) with a gap. -# Selected before any Cycle 7 detector look. Do not reassign after scoring. CYCLE7_EXPLORATORY_SEED_BASE = 810_000 CYCLE7_VALIDATION_SEED_BASE = 820_000 CYCLE7_CONFIRMATION_RESERVED_SEED_BASES = (830_000, 840_000, 850_000) diff --git a/fuckmark/cycle7/whitespace_collapse.py b/fuckmark/cycle7/whitespace_collapse.py index 43a4b3fc..25811185 100644 --- a/fuckmark/cycle7/whitespace_collapse.py +++ b/fuckmark/cycle7/whitespace_collapse.py @@ -27,17 +27,6 @@ def normalize_newlines(text: str) -> str: def collapse_horizontal_ascii_whitespace(text: str) -> str: - """Collapse runs of U+0020 and U+0009 to a single U+0020 per line. - - Specification (whitespace-collapse-v1): - - Convert CR LF and CR to LF first. - - Preserve LF as line structure. Do not join lines. - - On each line, replace every maximal run of SPACE (U+0020) and HT (U+0009) - with exactly one SPACE. - - Do not trim leftover single leading or trailing spaces. - - Do not collapse other Unicode whitespace (including NBSP U+00A0). - - Do not collapse vertical whitespace other than the CR/LF normalization above. - """ if not isinstance(text, str): raise TypeError("text must be a string") lines: list[str] = [] diff --git a/fuckmark/transforms/cycle7_quote_policy.py b/fuckmark/transforms/cycle7_quote_policy.py index d993373c..8744f7e1 100644 --- a/fuckmark/transforms/cycle7_quote_policy.py +++ b/fuckmark/transforms/cycle7_quote_policy.py @@ -1,7 +1,7 @@ from __future__ import annotations from collections.abc import Sequence -from typing import TYPE_CHECKING +from typing import TYPE_CHECKING, Any from .lexical_rules import LexicalTemplateRule from .quote_policy import ( @@ -76,13 +76,13 @@ def quote_interior_rule_allowed(quote_policy_id: str, rule: object) -> bool: def _quote_overlaps_for_operation( - quotes: Sequence[object], + quotes: Sequence[Any], operation: TransformOperation, -) -> tuple[object, ...]: +) -> tuple[Any, ...]: return tuple( span for span in quotes - if operation.source_start < span.end and span.start < operation.source_end # type: ignore[attr-defined] + if operation.source_start < span.end and span.start < operation.source_end ) From 4fcc6b2152dd4db65841b78f49b539fe958c397c Mon Sep 17 00:00:00 2001 From: byte271 Date: Tue, 25 Aug 2026 15:59:54 -0400 Subject: [PATCH 3/3] Add collapse-resistant compound and typographic-apostrophe families. Stage A on the same seed 810000 texts remains INSUFFICIENT_EVIDENCE: durable detections stay 4/4 after whitespace collapse. Co-authored-by: Cursor --- docs/cycle7/README.md | 27 ++-- docs/cycle7/durable-families.md | 90 +++++++++++++ docs/cycle7/literature.md | 18 ++- docs/cycle7/live-state-audit.md | 7 + docs/cycle7/stage-a-decision.md | 39 +++--- evidence/cycle7-stage-a-2026-08-25/README.md | 20 +-- .../cycle7-stage-a-2026-08-25/SHA256SUMS.txt | 8 +- .../detector-compare-family1-copula.json | 1 + .../detector-compare.json | 2 +- .../family2-density.json | 1 + .../fixture-geometry.json | 2 +- fuckmark/cycle7/density.py | 38 ++++++ fuckmark/cycle7/durable_rules.py | 61 ++++++++- fuckmark/cycle7/fixtures.py | 6 + fuckmark/cycle7/registry.py | 2 +- fuckmark/transforms/cycle7_quote_policy.py | 2 + fuckmark/transforms/lexical_rules.py | 40 +++++- tests/test_cycle7_apostrophe.py | 81 ++++++++++++ tests/test_cycle7_compounds.py | 124 ++++++++++++++++++ tests/test_cycle7_durable_rules.py | 9 +- 20 files changed, 520 insertions(+), 58 deletions(-) create mode 100644 docs/cycle7/durable-families.md create mode 100644 evidence/cycle7-stage-a-2026-08-25/detector-compare-family1-copula.json create mode 100644 evidence/cycle7-stage-a-2026-08-25/family2-density.json create mode 100644 fuckmark/cycle7/density.py create mode 100644 tests/test_cycle7_apostrophe.py create mode 100644 tests/test_cycle7_compounds.py diff --git a/docs/cycle7/README.md b/docs/cycle7/README.md index ff8f0a08..89d7285f 100644 --- a/docs/cycle7/README.md +++ b/docs/cycle7/README.md @@ -44,27 +44,30 @@ Cycle 7 evaluation arms: The Cycle 6 frozen sanitizer module is unchanged. -## Durable transform family (first attempt) +## Durable transform families -Unambiguous contractions (existing catalog + reversible have-forms) and UK/US `towards`/`toward`, `amongst`/`among`. +See `docs/cycle7/durable-families.md`. Catalog `cycle7-durable-rule-catalog-v2`. -These change tokens without relying on repeated U+0020. They survive ASCII whitespace collapse by construction. +1. Unambiguous contractions and closed orthography (Family 1). +2. Attested open ↔ hyphenated compounds (Family 2). +3. In-word ASCII apostrophe ↔ U+2019 (Family 3). + +These change tokens without relying on repeated U+0020. They survive ASCII whitespace collapse by construction. Family 3 also survives NFKC and Cf-strip. Quote interiors may receive these durable edits under `quote-container-durable-v1` without changing quote delimiters. Cycle 6 `quote-container-surface-spacing-v1` still admits only spacing inside quotes. ## Stage A result -**`INSUFFICIENT_EVIDENCE`** as a Cycle 6 replacement. - -Cycle 6 spacing: 0/4 raw detections, 4/4 after whitespace collapse (fresh seed `810000`). +Family 1 detector attachment on seed `810000` was **`INSUFFICIENT_EVIDENCE`**: Cycle 6 spacing 0/4 raw / 4/4 after collapse; durable 4/4 raw and 4/4 after collapse. -Cycle 7 durable family: 4/4 raw and 4/4 after collapse. Collapse-survival works on contraction-rich fixtures; natural density on GPT-2 TinyDev text is too low. +Families 2–3 raise candidate density on the same frozen texts (see `evidence/cycle7-stage-a-2026-08-25/family2-density.json`). Detector rescore of those texts under catalog v2 remains **`INSUFFICIENT_EVIDENCE`**: durable 4/4 raw and 4/4 after collapse; Cycle 6 spacing 0/4 raw and 4/4 after collapse; combined 1/4 raw and 4/4 after collapse. Details: `docs/cycle7/stage-a-decision.md`. +Not claimed: -- Not a Cycle 7 formal confirmation -- Not a ZERO_RESIDUAL claim -- Not human-fidelity validation -- Not threshold or FPR manipulation -- Not neural rewriting +- Cycle 7 formal confirmation +- ZERO_RESIDUAL +- Human-fidelity validation +- Threshold or FPR manipulation +- Neural rewriting diff --git a/docs/cycle7/durable-families.md b/docs/cycle7/durable-families.md new file mode 100644 index 00000000..b30a0582 --- /dev/null +++ b/docs/cycle7/durable-families.md @@ -0,0 +1,90 @@ +# Cycle 7 durable families + +Catalog identity: `cycle7-durable-rule-catalog-v2`. + +All families below are deterministic, detector-blind, key-blind, and non-neural. Cycle 6 `quote_safe_zrd` ruleset hash is unchanged. + +## Shared constraints + +- Protected spans (URLs, numbers, dates, code, identifiers, paths, hashes, citations, emails) block overlapping edits. +- Quote interiors: durable families may edit quoted *content* under `quote-container-durable-v1` without changing delimiters. Cycle 6 `quote-container-surface-spacing-v1` still admits only spacing inside quotes. +- Hard-invariant algorithm remains `hard-invariant-validator-v4`. Ambiguous `'d` / bare `it's` / `it's not` ↔ `it is not` stay excluded. +- Whitespace-collapse-v1 is a first-class Cycle 7 sanitizer. A family is collapse-resistant only if the rewrite still differs after `ws_collapse` and `ws_collapse_nfkc_cf_strip`. + +## Family 1 — unambiguous contractions and closed orthography + +Precondition: existing `LiteralTransformRule` whole-word match, all-caps blocked. + +Rewrite: closed contraction/expansion pairs plus `towards`/`toward` and `amongst`/`among`. Cycle 7 additions: `I/you/we/they have` ↔ `I've/you've/we've/they've`, and bounded `it's like|important|a|an|the` ↔ `it is …`. + +Inverse: each pair has an explicit reverse rule. + +Tokenizer: contraction boundaries change GPT-2 BPE pieces. VERIFIED on fixtures. + +Whitespace collapse: no extra U+0020 is introduced. Survives by construction. + +Sanitizer: ASCII punctuation; NFKC/Cf-strip leave the rewrite in place. + +Conflict geometry: overlapping contraction sites conflict; scheduler remains detector-blind cover-greedy v4. + +Enumeration order: registry sort `(start, end, rule_id, candidate_id)`. + +REJECTED from this family: `he'd`/`she'd`/`I'd`, bare `it's`, `it's not` ↔ `it is not`, `it's been`, `while`/`whilst` (noun `while` is not gated). + +## Family 2 — attested open ↔ hyphenated compounds + +Construction: `attested_open_hyphen_compound`. + +Closed list (both directions): + +- proof of concept ↔ proof-of-concept +- point of view ↔ point-of-view +- step by step ↔ step-by-step +- case by case ↔ case-by-case +- end to end ↔ end-to-end +- state of the art ↔ state-of-the-art +- face to face ↔ face-to-face + +Precondition: whole-word match, plus no adjacent hyphen, slash, or alnum so `proof-of-concept-note` and path fragments do not match. Sentence-final punctuation is allowed (unlike sentence-initial lexical markers). + +Rewrite: replace the matched open phrase with the hyphenated phrase, or the reverse. + +Inverse: the opposite direction in the same list. + +Tokenizer: GPT-2 `proof of concept` → 3 tokens; `proof-of-concept` → 5 tokens. VERIFIED locally on frozen GPT-2 revision `607a30d783dfa663caf39e06633721c8d4cfcd7e`. + +Whitespace collapse: hyphens are not spaces. Survives. + +Sanitizer: ASCII hyphens. NFKC/Cf-strip leave the rewrite in place. + +Conflict geometry: open and hyphenated forms of the same phrase cannot both match the same span. + +Enumeration order: same registry sort. + +Not admitted: `one to one` ↔ `one-to-one` (range false positive `one to one hundred` without extra gating). POS-gated pairs such as `set up`/`setup` and `real time`/`real-time` are not admitted. + +## Family 3 — in-word ASCII apostrophe ↔ U+2019 + +Construction: `inword_typographic_apostrophe`. + +Precondition: exactly one character `'` or U+2019 with alphabetic characters on both sides. Quote delimiters and plural possessives (`students'`) do not match. Both adjacent letters uppercase is rejected so `DON'T` does not fire. + +Rewrite: `'` → U+2019, or the reverse. + +Inverse: the opposite direction. + +Tokenizer: GPT-2 `It's not uncommon` tokens `[1026, 338, 407, 19185]` versus curly `[1026, 447, 247, 82, 407, 19185]`. VERIFIED on the frozen GPT-2 revision. + +Whitespace collapse: not whitespace. Survives. + +Sanitizer: U+2019 is category Pf, not Cf. NFKC leaves U+2019 unchanged. VERIFIED. Combined `ws_collapse_nfkc_cf_strip` therefore keeps the edit. NBSP and ZWSP were measured and rejected as channels: NBSP NFKC-maps to U+0020; ZWSP/soft hyphen are Cf. + +Hard invariants: `_WORD_RE` already treats `'` and U+2019 as equivalent after normalization. `don't` and `don’t` share the same negation atom. + +Conflict geometry: an apostrophe edit overlaps contraction expansion of the same word; cover-greedy v4 picks one. + +## Causal chain (what counts as promising) + +text change → tokenization change → n-gram/context change → observation replacement → g-value drift → score drop. + +Byte change or a one-off score drop is not enough. Family 1–3 have VERIFIED tokenization change and collapse survival on fixtures. Detector attachment on TinyDev 64-token text is a separate density question. diff --git a/docs/cycle7/literature.md b/docs/cycle7/literature.md index 5199a946..fd691e84 100644 --- a/docs/cycle7/literature.md +++ b/docs/cycle7/literature.md @@ -8,17 +8,29 @@ SOURCE-BOUND. Google documents SynthID-Text as a logits-processor watermark usin - https://ai.google.dev/responsible/docs/safeguards/synthid - https://huggingface.co/blog/synthid-text +- Dathathri et al., *Scalable watermarking for identifying large language model outputs*, Nature (2024): https://www.nature.com/articles/s41586-024-08025-4 This repository's Cycle 6 result is the local VERIFIED measurement for the frozen GPT-2 / mean-score protocol: 183/192 pristine detections reduced to 7/192 transformed detections, 0/192 transformed unwatermarked, `NONZERO_RESIDUAL`. That is not a claim about proprietary production detectors. ## Edit-channel / synchronization -SOURCE-BOUND / HYPOTHESIS. Context-hashed watermarks re-synchronize after a local edit once the sliding n-gram window has moved past the edit. Cycle 6 spacing inserts extra U+0020 tokens that change GPT-2 segmentation, then ordinary space-collapse removes those tokens and restores the original token sequence. That collapse failure is VERIFIED here as a mechanism class, not as a universal watermark theorem. +SOURCE-BOUND / HYPOTHESIS. Context-hashed watermarks re-synchronize after a local edit once the sliding n-gram window has moved past the edit. Cycle 6 spacing inserts extra U+0020 tokens that change GPT-2 segmentation, then ordinary space-collapse removes those tokens and restores the original token sequence. That collapse failure is VERIFIED here as a mechanism class on fresh seed `810000`, not as a universal watermark theorem. -EXTERNAL-VALIDATION-ONLY. Zhao et al. style analyses of SynthID robustness (paraphrase, back-translation, layer-inflation) are not reproduced in this repository and must not be cited as FuckMark results. +EXTERNAL-VALIDATION-ONLY. Zhao-style or layer-inflation analyses of SynthID, and synonym/paraphrase/back-translation robustness papers (including arXiv:2508.20228 and arXiv:2603.03410), are not reproduced in this repository and must not be cited as FuckMark results. Neural paraphrasing is out of protocol for the attack itself. ## Deterministic non-neural edits -HYPOTHESIS. Unambiguous contractions and closed orthographic variants change tokenizer boundaries without depending on repeated spaces. Whether they destroy enough SynthID observation windows on GPT-2 text is a Cycle 7 measurement, not a literature conclusion. +VERIFIED tokenizer effects on frozen GPT-2 `607a30d783dfa663caf39e06633721c8d4cfcd7e`: + +- `proof of concept` → 3 tokens; `proof-of-concept` → 5 tokens. +- ASCII `It's not uncommon` → `[1026, 338, 407, 19185]`; U+2019 apostrophe form → `[1026, 447, 247, 82, 407, 19185]`. + +VERIFIED sanitizer fate: + +- U+2019 survives NFKC and is not Cf, so it survives `ws_collapse_nfkc_cf_strip`. +- NBSP U+00A0 NFKC-maps to U+0020. REJECTED as a collapse-resistant channel. +- U+200B / U+00AD are Cf. REJECTED against the frozen Cf-strip arm. + +HYPOTHESIS. Unambiguous contractions, attested hyphenation, and typographic apostrophes can destroy some observation windows. Whether they destroy enough windows on 64-token TinyDev GPT-2 text is a Cycle 7 measurement. Family 1 detector attachment on seed `810000` was INSUFFICIENT_EVIDENCE because natural site density was too low. REJECTED as a Cycle 7 direction: adding more U+0020 spaces to chase 7/192 → 0/192 on spent confirmation rows. diff --git a/docs/cycle7/live-state-audit.md b/docs/cycle7/live-state-audit.md index f897b163..00e79518 100644 --- a/docs/cycle7/live-state-audit.md +++ b/docs/cycle7/live-state-audit.md @@ -55,3 +55,10 @@ SHA256SUMS.txt verified for every listed file. | Original freeze artifacts 760/770/780 | IDs and digests match the recovery manifest | Do not describe Cycle 6 as `ZERO_RESIDUAL`. + +## Continuation (2026-08-25, after PR #95 opened) + +Live GitHub still has `main` at `ddccd74a9e0e710085b385fed98959f2730b9d60`. Open PR: [#95](https://github.com/byte271/FuckMark/pull/95) on `research/cycle7-durable-transforms`. Open issues: none. + +Hygiene-only follow-up `6e587ea016cf5a2f4e7845a5550f97e0b5d92b34` made Release Readiness Baseline and CI succeed after `tests/test_repository_hygiene.py` rejected comments/docstrings in new Python. Cycle 6 sealed confirmation on that SHA succeeded. Codex review was usage-limited, not a code finding. CodeRabbit did not auto-review (repository star threshold). + diff --git a/docs/cycle7/stage-a-decision.md b/docs/cycle7/stage-a-decision.md index 0fe6c3c7..d1d63129 100644 --- a/docs/cycle7/stage-a-decision.md +++ b/docs/cycle7/stage-a-decision.md @@ -4,6 +4,8 @@ This is development-only. It is not a Cycle 7 formal confirmation. Seeds `830000` / `840000` / `850000` were not inspected. +Catalog `cycle7-durable-rule-catalog-v2` adds attested open/hyphen compounds and in-word typographic apostrophes on top of Family 1 contractions. The generated seed-`810000` texts were not regenerated. + ## What was measured Fresh exploratory seed `810000`, one pair per domain (8 texts), frozen GPT-2 revision, frozen Cycle 6 threshold `0.5570987654320988`, budget 14, detector-blind cover-greedy v4. @@ -11,42 +13,47 @@ Fresh exploratory seed `810000`, one pair per domain (8 texts), frozen GPT-2 rev Arms on the same texts: - Cycle 6 spacing-heavy B14 (`quote_safe_zrd_transform_registry`) -- Cycle 7 durable (contractions + bounded `it's` copulas + toward/among; no extra spaces) +- Cycle 7 durable (catalog v2; no extra spaces) - Combined durable + Cycle 6 spacing Sanitizer arms included `raw` and `ws_collapse`. ## Detector counts (watermarked 4, unwatermarked 4) -| Arm | Pristine WM | Raw WM | Raw UW | ws_collapse WM | ws_collapse UW | -| --- | --- | --- | --- | --- | --- | -| Cycle 6 spacing B14 | 4/4 | **0/4** | 0/4 | **4/4** | 0/4 | -| Cycle 7 durable | 4/4 | **4/4** | 0/4 | **4/4** | 0/4 | -| Combined | 4/4 | **0/4** | 0/4 | **4/4** | 0/4 | +Catalog v2 (latest `detector-compare.json`): + +| Arm | Pristine WM | Raw WM | Raw UW | ws_collapse WM | ws_collapse UW | Raw WM mean | +| --- | --- | --- | --- | --- | --- | --- | +| Cycle 6 spacing B14 | 4/4 | **0/4** | 0/4 | **4/4** | 0/4 | 0.525449 | +| Cycle 7 durable | 4/4 | **4/4** | 0/4 | **4/4** | 0/4 | 0.624884 | +| Combined | 4/4 | **1/4** | 0/4 | **4/4** | 0/4 | 0.531142 | -Family-1 snapshot (before bounded `it's` copulas) is preserved as `detector-compare-family1-pre-copula.json`. After adding `it's like` / `it's important` / `it's a|an|the`, durable still 4/4 detected. Two watermarked samples gained surviving non-space edits (`it's important`, `it's like`); that did not move detections under the frozen threshold. +Family 1 copula snapshot remains `detector-compare-family1-copula.json` (durable 4/4 raw and collapse; combined 0/4 raw, 4/4 collapse). ## VERIFIED (this Stage A) 1. Cycle 6 spacing can drive raw detections from 4/4 to 0/4 on this tiny corpus. -2. The same Cycle 6 outputs are detected 4/4 after `whitespace-collapse-v1`. Collapse restores the attack. This is the Cycle 6 channel weakness, reproduced on **new** development data, not on spent 760000–780000 rows. -3. Unwatermarked transformed controls stayed 0/4 on every arm. +2. The same Cycle 6 outputs are detected 4/4 after `whitespace-collapse-v1`. Collapse restores the attack. Reproduced on **new** development data, not on spent 760000–780000 rows. +3. Unwatermarked transformed controls stayed 0/4 on durable and Cycle 6 spacing. Combined unwatermarked stayed 0/4. 4. On contraction-rich fixtures, durable edits remain after collapse; inside quotes, Cycle 6 spacing collapses away and Cycle 7 durable edits remain. -5. Bare `it's not` ↔ `it is not` is not admitted: hard invariants treat `it's not` as a different negation atom than `is not`, matching the existing `he's not` / `he is not` split. +5. Bare `it's not` ↔ `it is not` is not admitted. +6. GPT-2 tokenization changes for Family 2 and Family 3 are real and collapse-resistant. Durable-only still leaves 4/4 detections because site density is too low (often 1 surviving non-space edit on a watermarked 64-token sample). Intact root windows remain ~44–53 of ~54–60. +7. Combined raw moved from 0/4 (Family 1) to 1/4 (catalog v2) on this corpus: cover-greedy spent a budget slot on `proof-of-concept` opening instead of a space, and that watermarked technical sample scored 0.5606 (threshold 0.5571). This is not used to retune the scheduler on these rows. It is a high-level geometry note: mixing sparse durable edits into a spacing budget can reduce raw coverage. ## SOURCE-BOUND forensic note (Cycle 6 spent residuals) Not used for tuning. High-level class only: most formal residuals looked like budget-limited leftover leak windows under a spacing-heavy attack. -## HYPOTHESIS going forward +## Decision -GPT-2 64-token TinyDev continuations rarely contain reversible contraction/orthography sites. A collapse-resistant family needs **higher natural density** than unambiguous copula/orthography, or a different tokenizer-boundary channel that is still non-neural and detector-blind. +**`INSUFFICIENT_EVIDENCE`** as a Cycle 6 replacement. Collapse-survival of the durable *channel* is real. Detector-side replacement of Cycle 6 spacing is not demonstrated on this 8-text exploratory corpus. -Do not chase 7/192 → 0/192 by adding more U+0020 spaces. +Do not open Cycle 7 formal confirmation. Do not chase 7/192 → 0/192 by adding U+0020 spaces. ## Next development direction 1. Keep Cycle 6 geometry/scheduler/protected-span machinery. -2. Search additional **high-frequency, invariant-safe** durable rules (not ambiguous `'d` / bare `it's`). -3. Measure density on Stage B seed `820000` only after a family is frozen. -4. Do not open Cycle 7 formal confirmation until a collapse-resistant family replicates on disjoint development data. +2. Search a durable family with **high natural density on ordinary TinyDev prose**, not only contraction/compound/apostrophe sites. Apostrophe-free and compound-free samples still have almost no durable candidates. +3. A detector-blind combined scheduler that fills spacing *after* durable ops is a hypothesis, not authorized from these four watermarked rows. +4. Measure Stage B seed `820000` only after a family is frozen. +5. REJECTED as durable channels: NBSP (NFKC→space), ZWSP/soft hyphen (Cf-strip). diff --git a/evidence/cycle7-stage-a-2026-08-25/README.md b/evidence/cycle7-stage-a-2026-08-25/README.md index 5419003a..746aa970 100644 --- a/evidence/cycle7-stage-a-2026-08-25/README.md +++ b/evidence/cycle7-stage-a-2026-08-25/README.md @@ -6,20 +6,10 @@ Files: - `fixture-geometry.json` — OffsetTokenizer geometry on hand fixtures, plus detector summaries when scored. - `decision-geometry-only.json` — classification from fixtures before detector scoring (`PROMISING_DEVELOPMENT` for the collapse-survival *channel*). -- `detector-compare-family1-pre-copula.json` — first durable catalog on the generated corpus (almost no sites). -- `detector-compare.json` — same generated texts rescored after bounded `it's` copulas. -- `decision.json` — Stage A decision after detector attachment: `INSUFFICIENT_EVIDENCE`. - -Do not treat these rows as confirmation data. Do not reuse them to retune Cycle 6. - - -Development-only. Not a formal confirmation. Seed `810000` is the exploratory ledger entry. - -Files: - -- `fixture-geometry.json` — OffsetTokenizer (byte) geometry on hand fixtures. CI-reproducible. -- `decision-geometry-only.json` — classification before detector scoring. -- `detector-compare.json` — GPT-2 / frozen threshold scoring on a fresh 8-sample corpus (added if the local HF run completes). -- `decision.json` — final Stage A decision after detector attachment when available. +- `detector-compare-family1-pre-copula.json` — Family 1 before bounded `it's` copulas (almost no sites). +- `detector-compare-family1-copula.json` — Family 1 after bounded copulas, same generated texts. +- `family2-density.json` — detector-blind candidate counts for catalog v2 on those same texts (no scores used for selection). +- `detector-compare.json` — latest detector rescore of the same generated texts under the current catalog. +- `decision.json` — Stage A decision after detector attachment. Do not treat these rows as confirmation data. Do not reuse them to retune Cycle 6. diff --git a/evidence/cycle7-stage-a-2026-08-25/SHA256SUMS.txt b/evidence/cycle7-stage-a-2026-08-25/SHA256SUMS.txt index 99004ae6..10b5d99d 100644 --- a/evidence/cycle7-stage-a-2026-08-25/SHA256SUMS.txt +++ b/evidence/cycle7-stage-a-2026-08-25/SHA256SUMS.txt @@ -1,6 +1,8 @@ 0fe1edfb1272911ba4a0052a67235edef58237483685acf21f122872e5a914b4 decision-geometry-only.json f2812647a9f5fcb13738947d6b613cf32af0893bf8c2a621c0c14e7273c2c1bf decision.json +4bf03dcdebdb39b52a62edb31dc53b811e53635bad1e4038a23e269f319ee760 detector-compare-family1-copula.json 84082cde1c2bf42911f637913dd0eb1e64f94529f1a1727dc4ad0b94fad7d63d detector-compare-family1-pre-copula.json -4bf03dcdebdb39b52a62edb31dc53b811e53635bad1e4038a23e269f319ee760 detector-compare.json -fc9bdc64db0639410ff3cfbaa18b6009c6e265a2913e4d04dab8361108c4eeff fixture-geometry.json -811257e314caedf3c94dfcafc3cc7943f0662e79b66f07a46b7c8cfa41e2a878 README.md +6bcb3dada456295b13822fa57e9e7eb1a398484c6614d57e3ee9765bacddd66f detector-compare.json +ec36359c4426f6691779874a5667c4ca1b4d810b4c9b230458653291386079f1 family2-density.json +1eb645110402bc67ce1ad181e4e66c14279394a75f4f152ed5721cd8d9be2297 fixture-geometry.json +833d1d94fcea01bc09855f0af3a4bcc18df1010065fb3e2825f8e2e31733c2eb README.md diff --git a/evidence/cycle7-stage-a-2026-08-25/detector-compare-family1-copula.json b/evidence/cycle7-stage-a-2026-08-25/detector-compare-family1-copula.json new file mode 100644 index 00000000..09e82894 --- /dev/null +++ b/evidence/cycle7-stage-a-2026-08-25/detector-compare-family1-copula.json @@ -0,0 +1 @@ +{"algorithm_version":"cycle7-stage-a-detector-compare-v1","artifact_hash":"4eaf4f34a855e7e46bd52ad2fc596bb1992a6639c483dc9958abe12e04808558","budget":14,"detector_access_used_for_selection":false,"durable_catalog_version":"cycle7-durable-rule-catalog-v1","geometry_rows":[{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"db8d7d1137950ec78c16db2aeba5ce6cb1fe2ef9d1f539bb04496aa0306d4b02","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"34484f77ef9a6549354b3b2baac224dce1cf6d0bb147445e0dadaaed77032d63","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":39,"measurement_hash":"3e7686ab59a4667f19eb7add5915410bad8c717cfea5dd7a5b2c636fbc8c8ea9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":39,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":39}},"domain":"general_explanatory","label":"unwatermarked","sample_id":"cycle7-stage-a-general_explanatory-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"54cd460f7640946d165a3d13cf2cf566e09f24c424f5eb16e30aa1768e57e2ed","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":59,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"6e8e88992a401f525e06faf4b74a5b68d12fc1bd33b97a40f2ae20b14137c1bd","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":55,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":55,"failure_classes":["candidate_exhaustion"],"intact_window_count":55,"measurement_hash":"9855d01e136f0355a45c2a1e5a274d7812b90eeb722e30e735cbc7e1412abb19","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-important"],"token_delta":0,"tuple_leak_window_count":55}},"domain":"general_explanatory","label":"watermarked","sample_id":"cycle7-stage-a-general_explanatory-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":4,"measurement_hash":"ea94965fd21cb116e8b37ecf1bf099857b9244d663d9e28feacc6f21e9632ad0","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-comma","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":7},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":59,"candidate_exhausted":false,"changed_token_positions":69,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":3,"measurement_hash":"63d019618be128cffadf6f2bfc7561ce8bb8b210ff783e2a0dd50e5be6591745","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","cycle7-expand-its-an","surface-space-after-any-word"],"token_delta":13,"tuple_leak_window_count":9},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["candidate_exhaustion"],"intact_window_count":35,"measurement_hash":"2a3fe2c19447deb585f7ccf6db0b11d339809cf3cda079734bc9229de5eeb98d","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":40,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-an"],"token_delta":0,"tuple_leak_window_count":35}},"domain":"technical_explanation","label":"unwatermarked","sample_id":"cycle7-stage-a-technical_explanation-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"a95ed6ff89d38bb3f97a1fe57ad8f8cd6b5f6af5fd667067773e41f6782bf373","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"93cb2a3afe25062bedd167067e8cf70c4c6e57e372fd7e6e30235670d5275db7","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":54,"measurement_hash":"7a5fd65de3dcdeb0322f31f88d1b24c796f6cc421d36494d9049403fe493a24a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":54,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":54}},"domain":"technical_explanation","label":"watermarked","sample_id":"cycle7-stage-a-technical_explanation-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":57,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":57,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":8,"measurement_hash":"93b4df80d55e224ac27b52995580fe92743ff437183dbcab8c52d613d64cf237","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":9},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":53,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":46,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":7,"measurement_hash":"aeaf328f84e96fddc16b96dd6850fe96a4ec58760c6bfb0fb9848992c98920ef","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","expand-you-are","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","expand-that-is-not-x","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":12,"tuple_leak_window_count":8},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":4,"candidate_exhausted":true,"changed_token_positions":5,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion"],"intact_window_count":40,"measurement_hash":"594f2148dd04a002541ac9bdd0a63a6aaeb1cb3fbc0e2a20dfbf6343d306146a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":57,"selected_count":3,"selected_rule_ids":["expand-you-are","expand-do-not","expand-that-is-not-x"],"token_delta":0,"tuple_leak_window_count":40}},"domain":"conversational_prose","label":"unwatermarked","sample_id":"cycle7-stage-a-conversational_prose-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"fa9c76c6ddb778141a4962a77e8002c1413ca7289d88c89177fe9d861683a4b5","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"3868bda9994b005506d125b7a70b2110f37b3b27cdb51c088ee19cd636b20c71","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"a815a53516b54ea1011be7ad4d6e8fc09b998a1bd7770f4c7171fad6431e4c4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59}},"domain":"conversational_prose","label":"watermarked","sample_id":"cycle7-stage-a-conversational_prose-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"c89d29f8c4ff8d8b907f91b255343f24aebb113f6f990e9dba00911566095e4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"9483f967307734bda93606f643cfad5f2ad1632797995b994ce16324a0dbfffc","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"d8e63ad3f8cddab8110faa1c91ecd15fc2161d622d41d3a655252553eb70c92a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59}},"domain":"structured_instructional","label":"unwatermarked","sample_id":"cycle7-stage-a-structured_instructional-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":44,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":13,"measurement_hash":"4fa592661bb841d8a40f7403a54f3c091faf666d726a680b733527f54d98a981","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":29,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma"],"token_delta":14,"tuple_leak_window_count":13},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":47,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":9,"measurement_hash":"26ce88d91c96c2cd4c2f98a4418a6b9d82afe975d7d9e9f897d1734cfb75c4af","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","cycle7-expand-its-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","expand-cannot","surface-space-after-any-word","cycle7-expand-i-have","surface-space-after-comma"],"token_delta":10,"tuple_leak_window_count":9},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":3,"candidate_exhausted":true,"changed_token_positions":18,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":44,"measurement_hash":"79aae0d2f1050d53e025cf30201df0f5f5cc6604cfd556a2384002c1df777cb9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":3,"selected_rule_ids":["cycle7-expand-its-like","expand-cannot","cycle7-expand-i-have"],"token_delta":-1,"tuple_leak_window_count":44}},"domain":"structured_instructional","label":"watermarked","sample_id":"cycle7-stage-a-structured_instructional-watermarked"}],"model":"openai-community/gpt2","model_revision":"607a30d783dfa663caf39e06633721c8d4cfcd7e","samples":[{"domain":"general_explanatory","label":"unwatermarked","sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","seed":810000,"text":"\n\nRead the documentation.\n\nWrite a proof of concept and a proof of proof of hypothesis.\n\nWrite a proof of concept and proof of hypothesis for a variety of reasons.\n\nWrite a proof of concept and proof of hypothesis for a variety of reasons. Create a proof of hypothesis and proof of hypothesis","text_sha256":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},{"domain":"general_explanatory","label":"watermarked","sample_id":"cycle7-stage-a-general_explanatory-watermarked","seed":810000,"text":" Use a scientific term for the same thing.\n\nHow do I keep track of my data?\n\nIt's not uncommon for someone to want to know if a new test set is working or not. I use a lot of tools and data sources to keep track of this information. It's important to keep your","text_sha256":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},{"domain":"technical_explanation","label":"unwatermarked","sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","seed":810033,"text":" Then, make a test suite. If the software fails, give the user a warning, but otherwise give the user an option to try again. If the software fails, give the user a warning, but otherwise give the user an option to try again. In a test suite, it's an important consideration that if the","text_sha256":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},{"domain":"technical_explanation","label":"watermarked","sample_id":"cycle7-stage-a-technical_explanation-watermarked","seed":810033,"text":" Check the validity of the hypothesis.\n\nTest for reproducibility\n\nFigure 3.5 (b) The following table summarizes the steps involved in reproducibility of the hypothesis.\n\n(a) Proof-of-concept (ROS)\n\nTo test reproducibility of the hypothesis, a","text_sha256":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},{"domain":"conversational_prose","label":"unwatermarked","sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","seed":810064,"text":"\n\nIn fact, when you're trying to answer a question, you may need to clarify the question to avoid confusion. Don't make a mistake that isn't relevant to the question.\n\nIn a nutshell:\n\nYou need to answer a question properly.\n\nYou need to give your answer in a","text_sha256":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"},{"domain":"conversational_prose","label":"watermarked","sample_id":"cycle7-stage-a-conversational_prose-watermarked","seed":810064,"text":"\n\nTry:\n\n1. Introduce a point of view, or question, in your own context.\n\n2. Discuss the importance of \"evidence\".\n\n3. Give a strong critique of your own research.\n\n4. Use the same words or phrases in your own work.\n\n","text_sha256":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},{"domain":"structured_instructional","label":"unwatermarked","sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","seed":810096,"text":"\n\nAn experimenter's description of the method is in the section below.\n\nThe test can be run on a computer or an operating system, using the \"C\" or \"F\" buttons. You can use the \"R\" or \"D\" buttons to control the program.\n\nThe program is","text_sha256":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},{"domain":"structured_instructional","label":"watermarked","sample_id":"cycle7-stage-a-structured_instructional-watermarked","seed":810096,"text":" Use a series of simple examples to illustrate the results.\n\n\"It's like having an elephant.\"\n\nA lot of people have asked me about this: \"How do you teach a learner to write a sentence that can't be changed at all?\"\n\nWhat I've learned, however, is that","text_sha256":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"}],"sanitizer_ids":["raw","nfkc","cf_strip","nfkc_cf_strip","ws_collapse","ws_collapse_nfkc_cf_strip"],"scored_rows":[{"arm_id":"cycle6_spacing_b14","domain":"general_explanatory","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"db8d7d1137950ec78c16db2aeba5ce6cb1fe2ef9d1f539bb04496aa0306d4b02","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc_cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"raw":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"ws_collapse":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"}}},{"arm_id":"cycle7_durable","domain":"general_explanatory","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":39,"measurement_hash":"3e7686ab59a4667f19eb7add5915410bad8c717cfea5dd7a5b2c636fbc8c8ea9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":39,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":39},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"nfkc":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"raw":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"}}},{"arm_id":"cycle7_combined","domain":"general_explanatory","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"34484f77ef9a6549354b3b2baac224dce1cf6d0bb147445e0dadaaed77032d63","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc_cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"raw":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"ws_collapse":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"}}},{"arm_id":"cycle6_spacing_b14","domain":"general_explanatory","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"54cd460f7640946d165a3d13cf2cf566e09f24c424f5eb16e30aa1768e57e2ed","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc_cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"raw":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"ws_collapse":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"}}},{"arm_id":"cycle7_durable","domain":"general_explanatory","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":55,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":55,"failure_classes":["candidate_exhaustion"],"intact_window_count":55,"measurement_hash":"9855d01e136f0355a45c2a1e5a274d7812b90eeb722e30e735cbc7e1412abb19","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-important"],"token_delta":0,"tuple_leak_window_count":55},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"nfkc":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"nfkc_cf_strip":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"raw":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"ws_collapse":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"}}},{"arm_id":"cycle7_combined","domain":"general_explanatory","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":59,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"6e8e88992a401f525e06faf4b74a5b68d12fc1bd33b97a40f2ae20b14137c1bd","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc_cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"raw":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"ws_collapse":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"}}},{"arm_id":"cycle6_spacing_b14","domain":"technical_explanation","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":4,"measurement_hash":"ea94965fd21cb116e8b37ecf1bf099857b9244d663d9e28feacc6f21e9632ad0","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-comma","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":7},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"nfkc":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"nfkc_cf_strip":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"raw":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"ws_collapse":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"}}},{"arm_id":"cycle7_durable","domain":"technical_explanation","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["candidate_exhaustion"],"intact_window_count":35,"measurement_hash":"2a3fe2c19447deb585f7ccf6db0b11d339809cf3cda079734bc9229de5eeb98d","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":40,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-an"],"token_delta":0,"tuple_leak_window_count":35},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"nfkc":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"nfkc_cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"raw":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"ws_collapse":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"}}},{"arm_id":"cycle7_combined","domain":"technical_explanation","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":59,"candidate_exhausted":false,"changed_token_positions":69,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":3,"measurement_hash":"63d019618be128cffadf6f2bfc7561ce8bb8b210ff783e2a0dd50e5be6591745","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","cycle7-expand-its-an","surface-space-after-any-word"],"token_delta":13,"tuple_leak_window_count":9},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"nfkc":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"nfkc_cf_strip":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"raw":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"ws_collapse":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"}}},{"arm_id":"cycle6_spacing_b14","domain":"technical_explanation","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"a95ed6ff89d38bb3f97a1fe57ad8f8cd6b5f6af5fd667067773e41f6782bf373","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc_cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"raw":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"ws_collapse":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"}}},{"arm_id":"cycle7_durable","domain":"technical_explanation","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":54,"measurement_hash":"7a5fd65de3dcdeb0322f31f88d1b24c796f6cc421d36494d9049403fe493a24a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":54,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":54},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"nfkc":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"raw":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"}}},{"arm_id":"cycle7_combined","domain":"technical_explanation","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"93cb2a3afe25062bedd167067e8cf70c4c6e57e372fd7e6e30235670d5275db7","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc_cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"raw":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"ws_collapse":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"}}},{"arm_id":"cycle6_spacing_b14","domain":"conversational_prose","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":57,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":57,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":8,"measurement_hash":"93b4df80d55e224ac27b52995580fe92743ff437183dbcab8c52d613d64cf237","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":9},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"nfkc":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"nfkc_cf_strip":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"raw":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"ws_collapse":{"detected":false,"score":0.5045631756158072,"text_hash":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5045631756158072,"text_hash":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"}}},{"arm_id":"cycle7_durable","domain":"conversational_prose","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":4,"candidate_exhausted":true,"changed_token_positions":5,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion"],"intact_window_count":40,"measurement_hash":"594f2148dd04a002541ac9bdd0a63a6aaeb1cb3fbc0e2a20dfbf6343d306146a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":57,"selected_count":3,"selected_rule_ids":["expand-you-are","expand-do-not","expand-that-is-not-x"],"token_delta":0,"tuple_leak_window_count":40},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"nfkc":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"nfkc_cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"raw":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"ws_collapse":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"}}},{"arm_id":"cycle7_combined","domain":"conversational_prose","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":53,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":46,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":7,"measurement_hash":"aeaf328f84e96fddc16b96dd6850fe96a4ec58760c6bfb0fb9848992c98920ef","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","expand-you-are","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","expand-that-is-not-x","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":12,"tuple_leak_window_count":8},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"nfkc":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"nfkc_cf_strip":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"raw":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"ws_collapse":{"detected":false,"score":0.4962785752259436,"text_hash":"9d238528d87390bd78adb434049836f4b4c913567ff043a539c90e5b4128c06d"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.4962785752259436,"text_hash":"9d238528d87390bd78adb434049836f4b4c913567ff043a539c90e5b4128c06d"}}},{"arm_id":"cycle6_spacing_b14","domain":"conversational_prose","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"fa9c76c6ddb778141a4962a77e8002c1413ca7289d88c89177fe9d861683a4b5","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc_cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"raw":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"ws_collapse":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"}}},{"arm_id":"cycle7_durable","domain":"conversational_prose","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"a815a53516b54ea1011be7ad4d6e8fc09b998a1bd7770f4c7171fad6431e4c4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"nfkc":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"raw":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"}}},{"arm_id":"cycle7_combined","domain":"conversational_prose","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"3868bda9994b005506d125b7a70b2110f37b3b27cdb51c088ee19cd636b20c71","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc_cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"raw":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"ws_collapse":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"}}},{"arm_id":"cycle6_spacing_b14","domain":"structured_instructional","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"c89d29f8c4ff8d8b907f91b255343f24aebb113f6f990e9dba00911566095e4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc_cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"raw":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle7_durable","domain":"structured_instructional","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"d8e63ad3f8cddab8110faa1c91ecd15fc2161d622d41d3a655252553eb70c92a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"nfkc":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"raw":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle7_combined","domain":"structured_instructional","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"9483f967307734bda93606f643cfad5f2ad1632797995b994ce16324a0dbfffc","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc_cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"raw":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle6_spacing_b14","domain":"structured_instructional","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":44,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":13,"measurement_hash":"4fa592661bb841d8a40f7403a54f3c091faf666d726a680b733527f54d98a981","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":29,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma"],"token_delta":14,"tuple_leak_window_count":13},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"nfkc":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"nfkc_cf_strip":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"raw":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"ws_collapse":{"detected":true,"score":0.653030303030303,"text_hash":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.653030303030303,"text_hash":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"}}},{"arm_id":"cycle7_durable","domain":"structured_instructional","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":3,"candidate_exhausted":true,"changed_token_positions":18,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":44,"measurement_hash":"79aae0d2f1050d53e025cf30201df0f5f5cc6604cfd556a2384002c1df777cb9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":3,"selected_rule_ids":["cycle7-expand-its-like","expand-cannot","cycle7-expand-i-have"],"token_delta":-1,"tuple_leak_window_count":44},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"nfkc":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"nfkc_cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"raw":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"ws_collapse":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"}}},{"arm_id":"cycle7_combined","domain":"structured_instructional","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":47,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":9,"measurement_hash":"26ce88d91c96c2cd4c2f98a4418a6b9d82afe975d7d9e9f897d1734cfb75c4af","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","cycle7-expand-its-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","expand-cannot","surface-space-after-any-word","cycle7-expand-i-have","surface-space-after-comma"],"token_delta":10,"tuple_leak_window_count":9},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"nfkc":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"nfkc_cf_strip":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"raw":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"ws_collapse":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"}}}],"secret_access_used_for_selection":false,"seed_base":810000,"summaries":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5254488114855487,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6426760741438237},"cycle7_combined":{"arm_id":"cycle7_combined","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5239653475020847,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6295426209679976},"cycle7_durable":{"arm_id":"cycle7_durable","median_raw_watermarked_score":0.6286616161616162,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":4,"raw_watermarked_mean_score":0.6275224189477956,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.6634137990070192,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6275224189477956}},"threshold":0.5570987654320988} diff --git a/evidence/cycle7-stage-a-2026-08-25/detector-compare.json b/evidence/cycle7-stage-a-2026-08-25/detector-compare.json index 09e82894..e9d38139 100644 --- a/evidence/cycle7-stage-a-2026-08-25/detector-compare.json +++ b/evidence/cycle7-stage-a-2026-08-25/detector-compare.json @@ -1 +1 @@ -{"algorithm_version":"cycle7-stage-a-detector-compare-v1","artifact_hash":"4eaf4f34a855e7e46bd52ad2fc596bb1992a6639c483dc9958abe12e04808558","budget":14,"detector_access_used_for_selection":false,"durable_catalog_version":"cycle7-durable-rule-catalog-v1","geometry_rows":[{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"db8d7d1137950ec78c16db2aeba5ce6cb1fe2ef9d1f539bb04496aa0306d4b02","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"34484f77ef9a6549354b3b2baac224dce1cf6d0bb147445e0dadaaed77032d63","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":39,"measurement_hash":"3e7686ab59a4667f19eb7add5915410bad8c717cfea5dd7a5b2c636fbc8c8ea9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":39,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":39}},"domain":"general_explanatory","label":"unwatermarked","sample_id":"cycle7-stage-a-general_explanatory-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"54cd460f7640946d165a3d13cf2cf566e09f24c424f5eb16e30aa1768e57e2ed","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":59,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"6e8e88992a401f525e06faf4b74a5b68d12fc1bd33b97a40f2ae20b14137c1bd","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":55,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":55,"failure_classes":["candidate_exhaustion"],"intact_window_count":55,"measurement_hash":"9855d01e136f0355a45c2a1e5a274d7812b90eeb722e30e735cbc7e1412abb19","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-important"],"token_delta":0,"tuple_leak_window_count":55}},"domain":"general_explanatory","label":"watermarked","sample_id":"cycle7-stage-a-general_explanatory-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":4,"measurement_hash":"ea94965fd21cb116e8b37ecf1bf099857b9244d663d9e28feacc6f21e9632ad0","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-comma","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":7},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":59,"candidate_exhausted":false,"changed_token_positions":69,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":3,"measurement_hash":"63d019618be128cffadf6f2bfc7561ce8bb8b210ff783e2a0dd50e5be6591745","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","cycle7-expand-its-an","surface-space-after-any-word"],"token_delta":13,"tuple_leak_window_count":9},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["candidate_exhaustion"],"intact_window_count":35,"measurement_hash":"2a3fe2c19447deb585f7ccf6db0b11d339809cf3cda079734bc9229de5eeb98d","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":40,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-an"],"token_delta":0,"tuple_leak_window_count":35}},"domain":"technical_explanation","label":"unwatermarked","sample_id":"cycle7-stage-a-technical_explanation-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"a95ed6ff89d38bb3f97a1fe57ad8f8cd6b5f6af5fd667067773e41f6782bf373","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"93cb2a3afe25062bedd167067e8cf70c4c6e57e372fd7e6e30235670d5275db7","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":54,"measurement_hash":"7a5fd65de3dcdeb0322f31f88d1b24c796f6cc421d36494d9049403fe493a24a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":54,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":54}},"domain":"technical_explanation","label":"watermarked","sample_id":"cycle7-stage-a-technical_explanation-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":57,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":57,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":8,"measurement_hash":"93b4df80d55e224ac27b52995580fe92743ff437183dbcab8c52d613d64cf237","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":9},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":53,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":46,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":7,"measurement_hash":"aeaf328f84e96fddc16b96dd6850fe96a4ec58760c6bfb0fb9848992c98920ef","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","expand-you-are","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","expand-that-is-not-x","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":12,"tuple_leak_window_count":8},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":4,"candidate_exhausted":true,"changed_token_positions":5,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion"],"intact_window_count":40,"measurement_hash":"594f2148dd04a002541ac9bdd0a63a6aaeb1cb3fbc0e2a20dfbf6343d306146a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":57,"selected_count":3,"selected_rule_ids":["expand-you-are","expand-do-not","expand-that-is-not-x"],"token_delta":0,"tuple_leak_window_count":40}},"domain":"conversational_prose","label":"unwatermarked","sample_id":"cycle7-stage-a-conversational_prose-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"fa9c76c6ddb778141a4962a77e8002c1413ca7289d88c89177fe9d861683a4b5","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"3868bda9994b005506d125b7a70b2110f37b3b27cdb51c088ee19cd636b20c71","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"a815a53516b54ea1011be7ad4d6e8fc09b998a1bd7770f4c7171fad6431e4c4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59}},"domain":"conversational_prose","label":"watermarked","sample_id":"cycle7-stage-a-conversational_prose-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"c89d29f8c4ff8d8b907f91b255343f24aebb113f6f990e9dba00911566095e4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"9483f967307734bda93606f643cfad5f2ad1632797995b994ce16324a0dbfffc","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"d8e63ad3f8cddab8110faa1c91ecd15fc2161d622d41d3a655252553eb70c92a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59}},"domain":"structured_instructional","label":"unwatermarked","sample_id":"cycle7-stage-a-structured_instructional-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":44,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":13,"measurement_hash":"4fa592661bb841d8a40f7403a54f3c091faf666d726a680b733527f54d98a981","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":29,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma"],"token_delta":14,"tuple_leak_window_count":13},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":47,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":9,"measurement_hash":"26ce88d91c96c2cd4c2f98a4418a6b9d82afe975d7d9e9f897d1734cfb75c4af","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","cycle7-expand-its-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","expand-cannot","surface-space-after-any-word","cycle7-expand-i-have","surface-space-after-comma"],"token_delta":10,"tuple_leak_window_count":9},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":3,"candidate_exhausted":true,"changed_token_positions":18,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":44,"measurement_hash":"79aae0d2f1050d53e025cf30201df0f5f5cc6604cfd556a2384002c1df777cb9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":3,"selected_rule_ids":["cycle7-expand-its-like","expand-cannot","cycle7-expand-i-have"],"token_delta":-1,"tuple_leak_window_count":44}},"domain":"structured_instructional","label":"watermarked","sample_id":"cycle7-stage-a-structured_instructional-watermarked"}],"model":"openai-community/gpt2","model_revision":"607a30d783dfa663caf39e06633721c8d4cfcd7e","samples":[{"domain":"general_explanatory","label":"unwatermarked","sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","seed":810000,"text":"\n\nRead the documentation.\n\nWrite a proof of concept and a proof of proof of hypothesis.\n\nWrite a proof of concept and proof of hypothesis for a variety of reasons.\n\nWrite a proof of concept and proof of hypothesis for a variety of reasons. Create a proof of hypothesis and proof of hypothesis","text_sha256":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},{"domain":"general_explanatory","label":"watermarked","sample_id":"cycle7-stage-a-general_explanatory-watermarked","seed":810000,"text":" Use a scientific term for the same thing.\n\nHow do I keep track of my data?\n\nIt's not uncommon for someone to want to know if a new test set is working or not. I use a lot of tools and data sources to keep track of this information. It's important to keep your","text_sha256":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},{"domain":"technical_explanation","label":"unwatermarked","sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","seed":810033,"text":" Then, make a test suite. If the software fails, give the user a warning, but otherwise give the user an option to try again. If the software fails, give the user a warning, but otherwise give the user an option to try again. In a test suite, it's an important consideration that if the","text_sha256":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},{"domain":"technical_explanation","label":"watermarked","sample_id":"cycle7-stage-a-technical_explanation-watermarked","seed":810033,"text":" Check the validity of the hypothesis.\n\nTest for reproducibility\n\nFigure 3.5 (b) The following table summarizes the steps involved in reproducibility of the hypothesis.\n\n(a) Proof-of-concept (ROS)\n\nTo test reproducibility of the hypothesis, a","text_sha256":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},{"domain":"conversational_prose","label":"unwatermarked","sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","seed":810064,"text":"\n\nIn fact, when you're trying to answer a question, you may need to clarify the question to avoid confusion. Don't make a mistake that isn't relevant to the question.\n\nIn a nutshell:\n\nYou need to answer a question properly.\n\nYou need to give your answer in a","text_sha256":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"},{"domain":"conversational_prose","label":"watermarked","sample_id":"cycle7-stage-a-conversational_prose-watermarked","seed":810064,"text":"\n\nTry:\n\n1. Introduce a point of view, or question, in your own context.\n\n2. Discuss the importance of \"evidence\".\n\n3. Give a strong critique of your own research.\n\n4. Use the same words or phrases in your own work.\n\n","text_sha256":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},{"domain":"structured_instructional","label":"unwatermarked","sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","seed":810096,"text":"\n\nAn experimenter's description of the method is in the section below.\n\nThe test can be run on a computer or an operating system, using the \"C\" or \"F\" buttons. You can use the \"R\" or \"D\" buttons to control the program.\n\nThe program is","text_sha256":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},{"domain":"structured_instructional","label":"watermarked","sample_id":"cycle7-stage-a-structured_instructional-watermarked","seed":810096,"text":" Use a series of simple examples to illustrate the results.\n\n\"It's like having an elephant.\"\n\nA lot of people have asked me about this: \"How do you teach a learner to write a sentence that can't be changed at all?\"\n\nWhat I've learned, however, is that","text_sha256":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"}],"sanitizer_ids":["raw","nfkc","cf_strip","nfkc_cf_strip","ws_collapse","ws_collapse_nfkc_cf_strip"],"scored_rows":[{"arm_id":"cycle6_spacing_b14","domain":"general_explanatory","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"db8d7d1137950ec78c16db2aeba5ce6cb1fe2ef9d1f539bb04496aa0306d4b02","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc_cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"raw":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"ws_collapse":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"}}},{"arm_id":"cycle7_durable","domain":"general_explanatory","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":39,"measurement_hash":"3e7686ab59a4667f19eb7add5915410bad8c717cfea5dd7a5b2c636fbc8c8ea9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":39,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":39},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"nfkc":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"raw":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"}}},{"arm_id":"cycle7_combined","domain":"general_explanatory","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"34484f77ef9a6549354b3b2baac224dce1cf6d0bb147445e0dadaaed77032d63","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc_cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"raw":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"ws_collapse":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"}}},{"arm_id":"cycle6_spacing_b14","domain":"general_explanatory","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"54cd460f7640946d165a3d13cf2cf566e09f24c424f5eb16e30aa1768e57e2ed","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc_cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"raw":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"ws_collapse":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"}}},{"arm_id":"cycle7_durable","domain":"general_explanatory","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":55,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":55,"failure_classes":["candidate_exhaustion"],"intact_window_count":55,"measurement_hash":"9855d01e136f0355a45c2a1e5a274d7812b90eeb722e30e735cbc7e1412abb19","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-important"],"token_delta":0,"tuple_leak_window_count":55},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"nfkc":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"nfkc_cf_strip":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"raw":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"ws_collapse":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6286616161616162,"text_hash":"a2920489751a7d0c43cb9723cd610051c31670c472cbfbe99692ff937382c137"}}},{"arm_id":"cycle7_combined","domain":"general_explanatory","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":59,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"6e8e88992a401f525e06faf4b74a5b68d12fc1bd33b97a40f2ae20b14137c1bd","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc_cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"raw":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"ws_collapse":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"}}},{"arm_id":"cycle6_spacing_b14","domain":"technical_explanation","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":4,"measurement_hash":"ea94965fd21cb116e8b37ecf1bf099857b9244d663d9e28feacc6f21e9632ad0","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-comma","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":7},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"nfkc":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"nfkc_cf_strip":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"raw":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"ws_collapse":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"}}},{"arm_id":"cycle7_durable","domain":"technical_explanation","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["candidate_exhaustion"],"intact_window_count":35,"measurement_hash":"2a3fe2c19447deb585f7ccf6db0b11d339809cf3cda079734bc9229de5eeb98d","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":40,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-an"],"token_delta":0,"tuple_leak_window_count":35},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"nfkc":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"nfkc_cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"raw":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"ws_collapse":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"}}},{"arm_id":"cycle7_combined","domain":"technical_explanation","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":59,"candidate_exhausted":false,"changed_token_positions":69,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":3,"measurement_hash":"63d019618be128cffadf6f2bfc7561ce8bb8b210ff783e2a0dd50e5be6591745","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","cycle7-expand-its-an","surface-space-after-any-word"],"token_delta":13,"tuple_leak_window_count":9},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"nfkc":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"nfkc_cf_strip":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"raw":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"ws_collapse":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"}}},{"arm_id":"cycle6_spacing_b14","domain":"technical_explanation","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"a95ed6ff89d38bb3f97a1fe57ad8f8cd6b5f6af5fd667067773e41f6782bf373","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc_cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"raw":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"ws_collapse":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"}}},{"arm_id":"cycle7_durable","domain":"technical_explanation","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":54,"measurement_hash":"7a5fd65de3dcdeb0322f31f88d1b24c796f6cc421d36494d9049403fe493a24a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":54,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":54},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"nfkc":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"raw":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"}}},{"arm_id":"cycle7_combined","domain":"technical_explanation","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"93cb2a3afe25062bedd167067e8cf70c4c6e57e372fd7e6e30235670d5275db7","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc_cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"raw":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"ws_collapse":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"}}},{"arm_id":"cycle6_spacing_b14","domain":"conversational_prose","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":57,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":57,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":8,"measurement_hash":"93b4df80d55e224ac27b52995580fe92743ff437183dbcab8c52d613d64cf237","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":9},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"nfkc":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"nfkc_cf_strip":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"raw":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"ws_collapse":{"detected":false,"score":0.5045631756158072,"text_hash":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5045631756158072,"text_hash":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"}}},{"arm_id":"cycle7_durable","domain":"conversational_prose","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":4,"candidate_exhausted":true,"changed_token_positions":5,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion"],"intact_window_count":40,"measurement_hash":"594f2148dd04a002541ac9bdd0a63a6aaeb1cb3fbc0e2a20dfbf6343d306146a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":57,"selected_count":3,"selected_rule_ids":["expand-you-are","expand-do-not","expand-that-is-not-x"],"token_delta":0,"tuple_leak_window_count":40},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"nfkc":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"nfkc_cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"raw":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"ws_collapse":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"}}},{"arm_id":"cycle7_combined","domain":"conversational_prose","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":53,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":46,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":7,"measurement_hash":"aeaf328f84e96fddc16b96dd6850fe96a4ec58760c6bfb0fb9848992c98920ef","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","expand-you-are","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","expand-that-is-not-x","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":12,"tuple_leak_window_count":8},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"nfkc":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"nfkc_cf_strip":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"raw":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"ws_collapse":{"detected":false,"score":0.4962785752259436,"text_hash":"9d238528d87390bd78adb434049836f4b4c913567ff043a539c90e5b4128c06d"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.4962785752259436,"text_hash":"9d238528d87390bd78adb434049836f4b4c913567ff043a539c90e5b4128c06d"}}},{"arm_id":"cycle6_spacing_b14","domain":"conversational_prose","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"fa9c76c6ddb778141a4962a77e8002c1413ca7289d88c89177fe9d861683a4b5","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc_cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"raw":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"ws_collapse":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"}}},{"arm_id":"cycle7_durable","domain":"conversational_prose","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"a815a53516b54ea1011be7ad4d6e8fc09b998a1bd7770f4c7171fad6431e4c4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"nfkc":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"raw":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"}}},{"arm_id":"cycle7_combined","domain":"conversational_prose","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"3868bda9994b005506d125b7a70b2110f37b3b27cdb51c088ee19cd636b20c71","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc_cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"raw":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"ws_collapse":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"}}},{"arm_id":"cycle6_spacing_b14","domain":"structured_instructional","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"c89d29f8c4ff8d8b907f91b255343f24aebb113f6f990e9dba00911566095e4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc_cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"raw":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle7_durable","domain":"structured_instructional","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":59,"measurement_hash":"d8e63ad3f8cddab8110faa1c91ecd15fc2161d622d41d3a655252553eb70c92a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":59},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"nfkc":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"raw":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle7_combined","domain":"structured_instructional","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"9483f967307734bda93606f643cfad5f2ad1632797995b994ce16324a0dbfffc","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc_cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"raw":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle6_spacing_b14","domain":"structured_instructional","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":44,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":13,"measurement_hash":"4fa592661bb841d8a40f7403a54f3c091faf666d726a680b733527f54d98a981","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":29,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma"],"token_delta":14,"tuple_leak_window_count":13},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"nfkc":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"nfkc_cf_strip":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"raw":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"ws_collapse":{"detected":true,"score":0.653030303030303,"text_hash":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.653030303030303,"text_hash":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"}}},{"arm_id":"cycle7_durable","domain":"structured_instructional","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":3,"candidate_exhausted":true,"changed_token_positions":18,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":44,"measurement_hash":"79aae0d2f1050d53e025cf30201df0f5f5cc6604cfd556a2384002c1df777cb9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":3,"selected_rule_ids":["cycle7-expand-its-like","expand-cannot","cycle7-expand-i-have"],"token_delta":-1,"tuple_leak_window_count":44},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"nfkc":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"nfkc_cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"raw":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"ws_collapse":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"}}},{"arm_id":"cycle7_combined","domain":"structured_instructional","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":47,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":9,"measurement_hash":"26ce88d91c96c2cd4c2f98a4418a6b9d82afe975d7d9e9f897d1734cfb75c4af","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","cycle7-expand-its-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","expand-cannot","surface-space-after-any-word","cycle7-expand-i-have","surface-space-after-comma"],"token_delta":10,"tuple_leak_window_count":9},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"nfkc":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"nfkc_cf_strip":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"raw":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"ws_collapse":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"}}}],"secret_access_used_for_selection":false,"seed_base":810000,"summaries":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5254488114855487,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6426760741438237},"cycle7_combined":{"arm_id":"cycle7_combined","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5239653475020847,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6295426209679976},"cycle7_durable":{"arm_id":"cycle7_durable","median_raw_watermarked_score":0.6286616161616162,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":4,"raw_watermarked_mean_score":0.6275224189477956,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.6634137990070192,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6275224189477956}},"threshold":0.5570987654320988} +{"algorithm_version":"cycle7-stage-a-detector-compare-v1","artifact_hash":"e98690d26117022f2b7cd31ca2c2e25b31960d18eb160dd2e1d67d5e0191eaa7","budget":14,"detector_access_used_for_selection":false,"durable_catalog_version":"cycle7-durable-rule-catalog-v2","geometry_rows":[{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"db8d7d1137950ec78c16db2aeba5ce6cb1fe2ef9d1f539bb04496aa0306d4b02","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":52,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":33,"collapsed_token_delta":2,"collapsed_tuple_leak_window_count":36,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":3,"measurement_hash":"99456740f5eff1de2b7d54e0bc047b6369bc38dd753527d12152d57ec7683360","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","lexical-compound-hyphenate-proof-of-concept","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":15,"tuple_leak_window_count":8},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":3,"candidate_exhausted":true,"changed_token_positions":54,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":31,"collapsed_token_delta":6,"collapsed_tuple_leak_window_count":31,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":31,"measurement_hash":"be518bc3941cb867c8f8ae545e7f9c489c2912b82ef5b9a6edb6019a859e8a7b","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":39,"selected_count":3,"selected_rule_ids":["lexical-compound-hyphenate-proof-of-concept","lexical-compound-hyphenate-proof-of-concept","lexical-compound-hyphenate-proof-of-concept"],"token_delta":6,"tuple_leak_window_count":31}},"domain":"general_explanatory","label":"unwatermarked","sample_id":"cycle7-stage-a-general_explanatory-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"54cd460f7640946d165a3d13cf2cf566e09f24c424f5eb16e30aa1768e57e2ed","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":61,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"a1a2f35c166f1187ed50d81ca1c0a45680748777f7045d56ee212e159835a995","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":45,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":3,"candidate_exhausted":true,"changed_token_positions":42,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":50,"collapsed_token_delta":2,"collapsed_tuple_leak_window_count":50,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":50,"measurement_hash":"43c49e5eb01ae92ebdf50c6dc513b873f02e181daa15f1c8023459fb48f4dde1","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":2,"selected_rule_ids":["lexical-apostrophe-ascii-to-typographic","cycle7-expand-its-important"],"token_delta":2,"tuple_leak_window_count":50}},"domain":"general_explanatory","label":"watermarked","sample_id":"cycle7-stage-a-general_explanatory-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":4,"measurement_hash":"ea94965fd21cb116e8b37ecf1bf099857b9244d663d9e28feacc6f21e9632ad0","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-comma","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":7},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":60,"candidate_exhausted":false,"changed_token_positions":69,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":3,"measurement_hash":"01651380ef912b68f7fb1f88bc8ffda902d28a31c41725446ddfedbcbec00df8","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","cycle7-expand-its-an","surface-space-after-any-word"],"token_delta":13,"tuple_leak_window_count":9},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":2,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["candidate_exhaustion"],"intact_window_count":35,"measurement_hash":"93044f331fcea9b50f3b0878d7d8936ace4aa923f91f51a761c4b5efa38754eb","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":40,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-an"],"token_delta":0,"tuple_leak_window_count":35}},"domain":"technical_explanation","label":"unwatermarked","sample_id":"cycle7-stage-a-technical_explanation-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"a95ed6ff89d38bb3f97a1fe57ad8f8cd6b5f6af5fd667067773e41f6782bf373","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":27,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":-2,"collapsed_tuple_leak_window_count":46,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":12,"measurement_hash":"5f5d096071fb0659708a67fbce0a1f29870a6ed1c433cddfc4af4c5126d5882c","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","lexical-compound-open-proof-of-concept","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":11,"tuple_leak_window_count":12},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":20,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":-2,"collapsed_tuple_leak_window_count":46,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":46,"measurement_hash":"1225dfeb8f3562e128fc2ea93bc534b9ca5089f5c255ea08fb3d22b68e79fa1f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":54,"selected_count":1,"selected_rule_ids":["lexical-compound-open-proof-of-concept"],"token_delta":-2,"tuple_leak_window_count":46}},"domain":"technical_explanation","label":"watermarked","sample_id":"cycle7-stage-a-technical_explanation-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":57,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":57,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":8,"measurement_hash":"93b4df80d55e224ac27b52995580fe92743ff437183dbcab8c52d613d64cf237","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":9},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":56,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":46,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":7,"measurement_hash":"b550b12d06cb5ec3734cdbda3cc5ce899a1182c3a556f1d9744977c6c53bd7b1","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","expand-you-are","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","expand-that-is-not-x","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":12,"tuple_leak_window_count":8},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":7,"candidate_exhausted":true,"changed_token_positions":5,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion"],"intact_window_count":40,"measurement_hash":"6b55f523f33aa7ac9479b879cd66e7ce93afdd8df642fc4c9936e34de1dfef29","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":57,"selected_count":3,"selected_rule_ids":["expand-you-are","expand-do-not","expand-that-is-not-x"],"token_delta":0,"tuple_leak_window_count":40}},"domain":"conversational_prose","label":"unwatermarked","sample_id":"cycle7-stage-a-conversational_prose-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"fa9c76c6ddb778141a4962a77e8002c1413ca7289d88c89177fe9d861683a4b5","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":44,"candidate_exhausted":false,"changed_token_positions":69,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":53,"collapsed_token_delta":2,"collapsed_tuple_leak_window_count":53,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":12,"measurement_hash":"fc0d1c5e9dc3938127c8af9b7c5fadc44f47f39c1635489686e517598807f513","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","lexical-compound-hyphenate-point-of-view","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":15,"tuple_leak_window_count":12},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":53,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":53,"collapsed_token_delta":2,"collapsed_tuple_leak_window_count":53,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":53,"measurement_hash":"1f7e7d49321e9300f4b100879eb97a9d9c6953f6bfde295d16192ba2ba60646d","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":1,"selected_rule_ids":["lexical-compound-hyphenate-point-of-view"],"token_delta":2,"tuple_leak_window_count":53}},"domain":"conversational_prose","label":"watermarked","sample_id":"cycle7-stage-a-conversational_prose-watermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"c89d29f8c4ff8d8b907f91b255343f24aebb113f6f990e9dba00911566095e4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":40,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"9655937528ef21a8f4bb9327aa3123d96fa7250614bca3ce07c752806e90767a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":26,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":61,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":54,"collapsed_token_delta":2,"collapsed_tuple_leak_window_count":54,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":54,"measurement_hash":"a20976821e4db5bb1c8eeb160863e843b9b79d1f710b4fe85986168eb86663e6","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":1,"selected_rule_ids":["lexical-apostrophe-ascii-to-typographic"],"token_delta":2,"tuple_leak_window_count":54}},"domain":"structured_instructional","label":"unwatermarked","sample_id":"cycle7-stage-a-structured_instructional-unwatermarked"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":44,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":13,"measurement_hash":"4fa592661bb841d8a40f7403a54f3c091faf666d726a680b733527f54d98a981","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":29,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma"],"token_delta":14,"tuple_leak_window_count":13},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":50,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":9,"measurement_hash":"8a62589faf940a247b8807578eba3f20aa58bb856deda05dcdc7d1999a89b789","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","cycle7-expand-its-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","expand-cannot","surface-space-after-any-word","cycle7-expand-i-have","surface-space-after-comma"],"token_delta":10,"tuple_leak_window_count":9},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":6,"candidate_exhausted":true,"changed_token_positions":18,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":44,"measurement_hash":"ce2302a8e6eb914fa5a29936ac0ed29571dadebd3037d737a2cbb1245b50f0a6","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":3,"selected_rule_ids":["cycle7-expand-its-like","expand-cannot","cycle7-expand-i-have"],"token_delta":-1,"tuple_leak_window_count":44}},"domain":"structured_instructional","label":"watermarked","sample_id":"cycle7-stage-a-structured_instructional-watermarked"}],"model":"openai-community/gpt2","model_revision":"607a30d783dfa663caf39e06633721c8d4cfcd7e","samples":[{"domain":"general_explanatory","label":"unwatermarked","sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","seed":810000,"text":"\n\nRead the documentation.\n\nWrite a proof of concept and a proof of proof of hypothesis.\n\nWrite a proof of concept and proof of hypothesis for a variety of reasons.\n\nWrite a proof of concept and proof of hypothesis for a variety of reasons. Create a proof of hypothesis and proof of hypothesis","text_sha256":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},{"domain":"general_explanatory","label":"watermarked","sample_id":"cycle7-stage-a-general_explanatory-watermarked","seed":810000,"text":" Use a scientific term for the same thing.\n\nHow do I keep track of my data?\n\nIt's not uncommon for someone to want to know if a new test set is working or not. I use a lot of tools and data sources to keep track of this information. It's important to keep your","text_sha256":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},{"domain":"technical_explanation","label":"unwatermarked","sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","seed":810033,"text":" Then, make a test suite. If the software fails, give the user a warning, but otherwise give the user an option to try again. If the software fails, give the user a warning, but otherwise give the user an option to try again. In a test suite, it's an important consideration that if the","text_sha256":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},{"domain":"technical_explanation","label":"watermarked","sample_id":"cycle7-stage-a-technical_explanation-watermarked","seed":810033,"text":" Check the validity of the hypothesis.\n\nTest for reproducibility\n\nFigure 3.5 (b) The following table summarizes the steps involved in reproducibility of the hypothesis.\n\n(a) Proof-of-concept (ROS)\n\nTo test reproducibility of the hypothesis, a","text_sha256":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},{"domain":"conversational_prose","label":"unwatermarked","sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","seed":810064,"text":"\n\nIn fact, when you're trying to answer a question, you may need to clarify the question to avoid confusion. Don't make a mistake that isn't relevant to the question.\n\nIn a nutshell:\n\nYou need to answer a question properly.\n\nYou need to give your answer in a","text_sha256":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"},{"domain":"conversational_prose","label":"watermarked","sample_id":"cycle7-stage-a-conversational_prose-watermarked","seed":810064,"text":"\n\nTry:\n\n1. Introduce a point of view, or question, in your own context.\n\n2. Discuss the importance of \"evidence\".\n\n3. Give a strong critique of your own research.\n\n4. Use the same words or phrases in your own work.\n\n","text_sha256":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},{"domain":"structured_instructional","label":"unwatermarked","sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","seed":810096,"text":"\n\nAn experimenter's description of the method is in the section below.\n\nThe test can be run on a computer or an operating system, using the \"C\" or \"F\" buttons. You can use the \"R\" or \"D\" buttons to control the program.\n\nThe program is","text_sha256":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},{"domain":"structured_instructional","label":"watermarked","sample_id":"cycle7-stage-a-structured_instructional-watermarked","seed":810096,"text":" Use a series of simple examples to illustrate the results.\n\n\"It's like having an elephant.\"\n\nA lot of people have asked me about this: \"How do you teach a learner to write a sentence that can't be changed at all?\"\n\nWhat I've learned, however, is that","text_sha256":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"}],"sanitizer_ids":["raw","nfkc","cf_strip","nfkc_cf_strip","ws_collapse","ws_collapse_nfkc_cf_strip"],"scored_rows":[{"arm_id":"cycle6_spacing_b14","domain":"general_explanatory","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":39,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":39,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":3,"measurement_hash":"db8d7d1137950ec78c16db2aeba5ce6cb1fe2ef9d1f539bb04496aa0306d4b02","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":10},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"nfkc_cf_strip":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"raw":{"detected":false,"score":0.5288853446748184,"text_hash":"9d2ec5efd08595f2cf1da68524a45c7f3baa0ef7a81164d1afe13e1ce54618cb"},"ws_collapse":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.48491323491323496,"text_hash":"1e20c029a8e678a063ad24e8a3840287506ef544a508f44b42c3fa90cd07ca28"}}},{"arm_id":"cycle7_durable","domain":"general_explanatory","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":3,"candidate_exhausted":true,"changed_token_positions":54,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":31,"collapsed_token_delta":6,"collapsed_tuple_leak_window_count":31,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":31,"measurement_hash":"be518bc3941cb867c8f8ae545e7f9c489c2912b82ef5b9a6edb6019a859e8a7b","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":39,"selected_count":3,"selected_rule_ids":["lexical-compound-hyphenate-proof-of-concept","lexical-compound-hyphenate-proof-of-concept","lexical-compound-hyphenate-proof-of-concept"],"token_delta":6,"tuple_leak_window_count":31},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.4482631189948263,"text_hash":"8bbde18e1217486da62183d11d777afaeb48342037509eee0333f6959f49c54d"},"nfkc":{"detected":false,"score":0.4482631189948263,"text_hash":"8bbde18e1217486da62183d11d777afaeb48342037509eee0333f6959f49c54d"},"nfkc_cf_strip":{"detected":false,"score":0.4482631189948263,"text_hash":"8bbde18e1217486da62183d11d777afaeb48342037509eee0333f6959f49c54d"},"raw":{"detected":false,"score":0.4482631189948263,"text_hash":"8bbde18e1217486da62183d11d777afaeb48342037509eee0333f6959f49c54d"},"ws_collapse":{"detected":false,"score":0.4482631189948263,"text_hash":"8bbde18e1217486da62183d11d777afaeb48342037509eee0333f6959f49c54d"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.4482631189948263,"text_hash":"8bbde18e1217486da62183d11d777afaeb48342037509eee0333f6959f49c54d"}}},{"arm_id":"cycle7_combined","domain":"general_explanatory","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":52,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":33,"collapsed_token_delta":2,"collapsed_tuple_leak_window_count":36,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":3,"measurement_hash":"99456740f5eff1de2b7d54e0bc047b6369bc38dd753527d12152d57ec7683360","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":39,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","lexical-compound-hyphenate-proof-of-concept","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":15,"tuple_leak_window_count":8},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.48491323491323496,"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5179763739085773,"text_hash":"fcfcaf3c6ae94d66beb89e35065f47d833e2886288c50258d5d45ea13a67754d"},"nfkc":{"detected":false,"score":0.5179763739085773,"text_hash":"fcfcaf3c6ae94d66beb89e35065f47d833e2886288c50258d5d45ea13a67754d"},"nfkc_cf_strip":{"detected":false,"score":0.5179763739085773,"text_hash":"fcfcaf3c6ae94d66beb89e35065f47d833e2886288c50258d5d45ea13a67754d"},"raw":{"detected":false,"score":0.5179763739085773,"text_hash":"fcfcaf3c6ae94d66beb89e35065f47d833e2886288c50258d5d45ea13a67754d"},"ws_collapse":{"detected":false,"score":0.4575298438934803,"text_hash":"469940cad6bc155a715ee0b0869f2a327e37250f1696a98bcc9d6edb2d0a61ee"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.4575298438934803,"text_hash":"469940cad6bc155a715ee0b0869f2a327e37250f1696a98bcc9d6edb2d0a61ee"}}},{"arm_id":"cycle6_spacing_b14","domain":"general_explanatory","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"54cd460f7640946d165a3d13cf2cf566e09f24c424f5eb16e30aa1768e57e2ed","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc_cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"raw":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"ws_collapse":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"}}},{"arm_id":"cycle7_durable","domain":"general_explanatory","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":3,"candidate_exhausted":true,"changed_token_positions":42,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":50,"collapsed_token_delta":2,"collapsed_tuple_leak_window_count":50,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":50,"measurement_hash":"43c49e5eb01ae92ebdf50c6dc513b873f02e181daa15f1c8023459fb48f4dde1","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":2,"selected_rule_ids":["lexical-apostrophe-ascii-to-typographic","cycle7-expand-its-important"],"token_delta":2,"tuple_leak_window_count":50},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6248778103616813,"text_hash":"959a99c03f543a2d5327e82e1b87d35dba03366161223c300d07550c3c119e98"},"nfkc":{"detected":true,"score":0.6248778103616813,"text_hash":"959a99c03f543a2d5327e82e1b87d35dba03366161223c300d07550c3c119e98"},"nfkc_cf_strip":{"detected":true,"score":0.6248778103616813,"text_hash":"959a99c03f543a2d5327e82e1b87d35dba03366161223c300d07550c3c119e98"},"raw":{"detected":true,"score":0.6248778103616813,"text_hash":"959a99c03f543a2d5327e82e1b87d35dba03366161223c300d07550c3c119e98"},"ws_collapse":{"detected":true,"score":0.6248778103616813,"text_hash":"959a99c03f543a2d5327e82e1b87d35dba03366161223c300d07550c3c119e98"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6248778103616813,"text_hash":"959a99c03f543a2d5327e82e1b87d35dba03366161223c300d07550c3c119e98"}}},{"arm_id":"cycle7_combined","domain":"general_explanatory","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":61,"candidate_exhausted":false,"changed_token_positions":75,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"a1a2f35c166f1187ed50d81ca1c0a45680748777f7045d56ee212e159835a995","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":45,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-keep","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period"],"token_delta":14,"tuple_leak_window_count":12},"label":"watermarked","pristine_detected":true,"pristine_score":0.6367424242424242,"sample_id":"cycle7-stage-a-general_explanatory-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"nfkc_cf_strip":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"raw":{"detected":false,"score":0.5242970242970243,"text_hash":"734b31f00f9223c01eaa78f7440361fc3d2dfba58b76d4d456e393f4af080836"},"ws_collapse":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6367424242424242,"text_hash":"af6ca3b75d2c085cf82de93660111d0a587a321ebe559f6d781d64f8f2454f5b"}}},{"arm_id":"cycle6_spacing_b14","domain":"technical_explanation","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":58,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":4,"measurement_hash":"ea94965fd21cb116e8b37ecf1bf099857b9244d663d9e28feacc6f21e9632ad0","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":43,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-comma","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":7},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"nfkc":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"nfkc_cf_strip":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"raw":{"detected":false,"score":0.48607038123167157,"text_hash":"9f8ed5704aaf940f8a3e8aacf639d764bf8f93f72ffc3fbad905cadaec4c1dd5"},"ws_collapse":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5303030303030303,"text_hash":"e6c2764c43a545455595108822987cb7b839026211b21240f31d585da11e8ecc"}}},{"arm_id":"cycle7_durable","domain":"technical_explanation","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":2,"candidate_exhausted":true,"changed_token_positions":1,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["candidate_exhaustion"],"intact_window_count":35,"measurement_hash":"93044f331fcea9b50f3b0878d7d8936ace4aa923f91f51a761c4b5efa38754eb","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":40,"selected_count":1,"selected_rule_ids":["cycle7-expand-its-an"],"token_delta":0,"tuple_leak_window_count":35},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"nfkc":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"nfkc_cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"raw":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"ws_collapse":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"}}},{"arm_id":"cycle7_combined","domain":"technical_explanation","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":60,"candidate_exhausted":false,"changed_token_positions":69,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":35,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":35,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":3,"measurement_hash":"01651380ef912b68f7fb1f88bc8ffda902d28a31c41725446ddfedbcbec00df8","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":42,"root_window_count":40,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","cycle7-expand-its-an","surface-space-after-any-word"],"token_delta":13,"tuple_leak_window_count":9},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5303030303030303,"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"nfkc":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"nfkc_cf_strip":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"raw":{"detected":false,"score":0.48690292758089365,"text_hash":"b7a5adc44aa363e8e0a49d7464c2c624dad7c75f2196af3f7bb0a0c0d7776090"},"ws_collapse":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5230429292929293,"text_hash":"b8ab13559be295b49615e440a615da753e65a26fca5285d9d5aa0a0f4a41f5be"}}},{"arm_id":"cycle6_spacing_b14","domain":"technical_explanation","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":54,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":54,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":16,"measurement_hash":"a95ed6ff89d38bb3f97a1fe57ad8f8cd6b5f6af5fd667067773e41f6782bf373","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":16},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"nfkc_cf_strip":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"raw":{"detected":false,"score":0.5473293541475359,"text_hash":"597b0ebb33b4564b65cf9096ea6313d2a173073083c72fa3f1651cc5173b4317"},"ws_collapse":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.617517770295548,"text_hash":"9c89b6ed52d5e5e9294624b301aa89507b48f93edc6d11e0f7c81d2a696d43ed"}}},{"arm_id":"cycle7_durable","domain":"technical_explanation","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":20,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":-2,"collapsed_tuple_leak_window_count":46,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":46,"measurement_hash":"1225dfeb8f3562e128fc2ea93bc534b9ca5089f5c255ea08fb3d22b68e79fa1f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":54,"selected_count":1,"selected_rule_ids":["lexical-compound-open-proof-of-concept"],"token_delta":-2,"tuple_leak_window_count":46},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6214063714063714,"text_hash":"bb1195a06da4c1944b77dd18fa8d68360ec411f67645a775d7ddd2ecdd3d2c67"},"nfkc":{"detected":true,"score":0.6214063714063714,"text_hash":"bb1195a06da4c1944b77dd18fa8d68360ec411f67645a775d7ddd2ecdd3d2c67"},"nfkc_cf_strip":{"detected":true,"score":0.6214063714063714,"text_hash":"bb1195a06da4c1944b77dd18fa8d68360ec411f67645a775d7ddd2ecdd3d2c67"},"raw":{"detected":true,"score":0.6214063714063714,"text_hash":"bb1195a06da4c1944b77dd18fa8d68360ec411f67645a775d7ddd2ecdd3d2c67"},"ws_collapse":{"detected":true,"score":0.6214063714063714,"text_hash":"bb1195a06da4c1944b77dd18fa8d68360ec411f67645a775d7ddd2ecdd3d2c67"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6214063714063714,"text_hash":"bb1195a06da4c1944b77dd18fa8d68360ec411f67645a775d7ddd2ecdd3d2c67"}}},{"arm_id":"cycle7_combined","domain":"technical_explanation","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":27,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":-2,"collapsed_tuple_leak_window_count":46,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":12,"measurement_hash":"5f5d096071fb0659708a67fbce0a1f29870a6ed1c433cddfc4af4c5126d5882c","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":12,"root_window_count":54,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","lexical-compound-open-proof-of-concept","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":11,"tuple_leak_window_count":12},"label":"watermarked","pristine_detected":true,"pristine_score":0.617517770295548,"sample_id":"cycle7-stage-a-technical_explanation-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.5606060606060607,"text_hash":"5a0a39e4201d5b1a7dbb64fe828e3c6b5b9d1603ca789c0d02385c3c95deb560"},"nfkc":{"detected":true,"score":0.5606060606060607,"text_hash":"5a0a39e4201d5b1a7dbb64fe828e3c6b5b9d1603ca789c0d02385c3c95deb560"},"nfkc_cf_strip":{"detected":true,"score":0.5606060606060607,"text_hash":"5a0a39e4201d5b1a7dbb64fe828e3c6b5b9d1603ca789c0d02385c3c95deb560"},"raw":{"detected":true,"score":0.5606060606060607,"text_hash":"5a0a39e4201d5b1a7dbb64fe828e3c6b5b9d1603ca789c0d02385c3c95deb560"},"ws_collapse":{"detected":true,"score":0.6214063714063714,"text_hash":"bb1195a06da4c1944b77dd18fa8d68360ec411f67645a775d7ddd2ecdd3d2c67"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6214063714063714,"text_hash":"bb1195a06da4c1944b77dd18fa8d68360ec411f67645a775d7ddd2ecdd3d2c67"}}},{"arm_id":"cycle6_spacing_b14","domain":"conversational_prose","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":49,"candidate_exhausted":false,"changed_token_positions":71,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":57,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":57,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":8,"measurement_hash":"93b4df80d55e224ac27b52995580fe92743ff437183dbcab8c52d613d64cf237","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":34,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":9},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"nfkc":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"nfkc_cf_strip":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"raw":{"detected":false,"score":0.4962121212121212,"text_hash":"e22706a24a97c1dfae7806af1cdcd011ce2c4bd7110c090622d2f26250dceea1"},"ws_collapse":{"detected":false,"score":0.5045631756158072,"text_hash":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5045631756158072,"text_hash":"2f23b386de94e16f057c68b2fbe8ba46bd1aca05f3a202d278cd606a407030ea"}}},{"arm_id":"cycle7_durable","domain":"conversational_prose","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":7,"candidate_exhausted":true,"changed_token_positions":5,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion"],"intact_window_count":40,"measurement_hash":"6b55f523f33aa7ac9479b879cd66e7ce93afdd8df642fc4c9936e34de1dfef29","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":57,"selected_count":3,"selected_rule_ids":["expand-you-are","expand-do-not","expand-that-is-not-x"],"token_delta":0,"tuple_leak_window_count":40},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"nfkc":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"nfkc_cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"raw":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"ws_collapse":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.49743044479886583,"text_hash":"32a004f3aa3fc1850c3625ecd6696e5ce0ff37b907f414b68d9a993db02a9240"}}},{"arm_id":"cycle7_combined","domain":"conversational_prose","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":56,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":46,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":46,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":7,"measurement_hash":"b550b12d06cb5ec3734cdbda3cc5ce899a1182c3a556f1d9744977c6c53bd7b1","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":35,"root_window_count":57,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","expand-you-are","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","expand-that-is-not-x","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":12,"tuple_leak_window_count":8},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5045631756158072,"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"nfkc":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"nfkc_cf_strip":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"raw":{"detected":false,"score":0.4858443590837957,"text_hash":"b57a62afcb37a292d527b0676efd8778c4f94573da8d7b943fcb370b4feaee75"},"ws_collapse":{"detected":false,"score":0.4962785752259436,"text_hash":"9d238528d87390bd78adb434049836f4b4c913567ff043a539c90e5b4128c06d"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.4962785752259436,"text_hash":"9d238528d87390bd78adb434049836f4b4c913567ff043a539c90e5b4128c06d"}}},{"arm_id":"cycle6_spacing_b14","domain":"conversational_prose","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":43,"candidate_exhausted":false,"changed_token_positions":67,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":14,"measurement_hash":"fa9c76c6ddb778141a4962a77e8002c1413ca7289d88c89177fe9d861683a4b5","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-comma","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":14},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"nfkc_cf_strip":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"raw":{"detected":false,"score":0.5178497301784973,"text_hash":"44314450c696e31576ad2b42ae40437c9b9d30379852b7a144b4f6e3f8602205"},"ws_collapse":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6634137990070192,"text_hash":"93b150afa11b41128b6f41950a305b9f74d7723c0dcbbb066ffcc4a2060922c9"}}},{"arm_id":"cycle7_durable","domain":"conversational_prose","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":53,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":53,"collapsed_token_delta":2,"collapsed_tuple_leak_window_count":53,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":53,"measurement_hash":"1f7e7d49321e9300f4b100879eb97a9d9c6953f6bfde295d16192ba2ba60646d","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":1,"selected_rule_ids":["lexical-compound-hyphenate-point-of-view"],"token_delta":2,"tuple_leak_window_count":53},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6527570789865872,"text_hash":"2f57624b36fb5514110b348173204292136a51c9e101e5ed9ddb32cfd7e7adad"},"nfkc":{"detected":true,"score":0.6527570789865872,"text_hash":"2f57624b36fb5514110b348173204292136a51c9e101e5ed9ddb32cfd7e7adad"},"nfkc_cf_strip":{"detected":true,"score":0.6527570789865872,"text_hash":"2f57624b36fb5514110b348173204292136a51c9e101e5ed9ddb32cfd7e7adad"},"raw":{"detected":true,"score":0.6527570789865872,"text_hash":"2f57624b36fb5514110b348173204292136a51c9e101e5ed9ddb32cfd7e7adad"},"ws_collapse":{"detected":true,"score":0.6527570789865872,"text_hash":"2f57624b36fb5514110b348173204292136a51c9e101e5ed9ddb32cfd7e7adad"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6527570789865872,"text_hash":"2f57624b36fb5514110b348173204292136a51c9e101e5ed9ddb32cfd7e7adad"}}},{"arm_id":"cycle7_combined","domain":"conversational_prose","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":44,"candidate_exhausted":false,"changed_token_positions":69,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":53,"collapsed_token_delta":2,"collapsed_tuple_leak_window_count":53,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":12,"measurement_hash":"fc0d1c5e9dc3938127c8af9b7c5fadc44f47f39c1635489686e517598807f513","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-period","lexical-compound-hyphenate-point-of-view","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":15,"tuple_leak_window_count":12},"label":"watermarked","pristine_detected":true,"pristine_score":0.6634137990070192,"sample_id":"cycle7-stage-a-conversational_prose-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5332779853327798,"text_hash":"445d176ef884334971f2f32feb6a90fbfc67f98574e1da8e7ccdf5ba6ebd4650"},"nfkc":{"detected":false,"score":0.5332779853327798,"text_hash":"445d176ef884334971f2f32feb6a90fbfc67f98574e1da8e7ccdf5ba6ebd4650"},"nfkc_cf_strip":{"detected":false,"score":0.5332779853327798,"text_hash":"445d176ef884334971f2f32feb6a90fbfc67f98574e1da8e7ccdf5ba6ebd4650"},"raw":{"detected":false,"score":0.5332779853327798,"text_hash":"445d176ef884334971f2f32feb6a90fbfc67f98574e1da8e7ccdf5ba6ebd4650"},"ws_collapse":{"detected":true,"score":0.6527570789865872,"text_hash":"2f57624b36fb5514110b348173204292136a51c9e101e5ed9ddb32cfd7e7adad"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6527570789865872,"text_hash":"2f57624b36fb5514110b348173204292136a51c9e101e5ed9ddb32cfd7e7adad"}}},{"arm_id":"cycle6_spacing_b14","domain":"structured_instructional","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":39,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"c89d29f8c4ff8d8b907f91b255343f24aebb113f6f990e9dba00911566095e4f","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":25,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc_cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"raw":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle7_durable","domain":"structured_instructional","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":1,"candidate_exhausted":true,"changed_token_positions":61,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":54,"collapsed_token_delta":2,"collapsed_tuple_leak_window_count":54,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":54,"measurement_hash":"a20976821e4db5bb1c8eeb160863e843b9b79d1f710b4fe85986168eb86663e6","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":59,"selected_count":1,"selected_rule_ids":["lexical-apostrophe-ascii-to-typographic"],"token_delta":2,"tuple_leak_window_count":54},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5110531544957774,"text_hash":"550e85e43f030f15f9ed39cac28b51012b0e543904986dd39aa667fc5463f452"},"nfkc":{"detected":false,"score":0.5110531544957774,"text_hash":"550e85e43f030f15f9ed39cac28b51012b0e543904986dd39aa667fc5463f452"},"nfkc_cf_strip":{"detected":false,"score":0.5110531544957774,"text_hash":"550e85e43f030f15f9ed39cac28b51012b0e543904986dd39aa667fc5463f452"},"raw":{"detected":false,"score":0.5110531544957774,"text_hash":"550e85e43f030f15f9ed39cac28b51012b0e543904986dd39aa667fc5463f452"},"ws_collapse":{"detected":false,"score":0.5110531544957774,"text_hash":"550e85e43f030f15f9ed39cac28b51012b0e543904986dd39aa667fc5463f452"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5110531544957774,"text_hash":"550e85e43f030f15f9ed39cac28b51012b0e543904986dd39aa667fc5463f452"}}},{"arm_id":"cycle7_combined","domain":"structured_instructional","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":40,"candidate_exhausted":false,"changed_token_positions":72,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":59,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":59,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":12,"measurement_hash":"9655937528ef21a8f4bb9327aa3123d96fa7250614bca3ce07c752806e90767a","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":26,"root_window_count":59,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-period","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":12},"label":"unwatermarked","pristine_detected":false,"pristine_score":0.5104434172230783,"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"nfkc_cf_strip":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"raw":{"detected":false,"score":0.5075056116722784,"text_hash":"5b637aabb37624c9f3d9cd5314e2a6594282ed30f5eb44ef4eeceb69cfc46435"},"ws_collapse":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"},"ws_collapse_nfkc_cf_strip":{"detected":false,"score":0.5104434172230783,"text_hash":"598aac9fc26b16716017ed531bd6657ed293c4c8a26a9efda293bbb86a0dfdd4"}}},{"arm_id":"cycle6_spacing_b14","domain":"structured_instructional","geometry":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":44,"candidate_exhausted":false,"changed_token_positions":74,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":60,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":60,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":13,"measurement_hash":"4fa592661bb841d8a40f7403a54f3c091faf666d726a680b733527f54d98a981","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":29,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-comma"],"token_delta":14,"tuple_leak_window_count":13},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"nfkc":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"nfkc_cf_strip":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"raw":{"detected":false,"score":0.5123191373191373,"text_hash":"258c93499a81d0ebb30368fbce0273d4c32f686a41a5039c8b04e82c52999494"},"ws_collapse":{"detected":true,"score":0.653030303030303,"text_hash":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.653030303030303,"text_hash":"d4675e31435c61befbed5ded806f205fe3f629bceb162cdce3ffffff2c8bf7a9"}}},{"arm_id":"cycle7_durable","domain":"structured_instructional","geometry":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":6,"candidate_exhausted":true,"changed_token_positions":18,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":44,"measurement_hash":"ce2302a8e6eb914fa5a29936ac0ed29571dadebd3037d737a2cbb1245b50f0a6","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":60,"selected_count":3,"selected_rule_ids":["cycle7-expand-its-like","expand-cannot","cycle7-expand-i-have"],"token_delta":-1,"tuple_leak_window_count":44},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"nfkc":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"nfkc_cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"raw":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"ws_collapse":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"}}},{"arm_id":"cycle7_combined","domain":"structured_instructional","geometry":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":50,"candidate_exhausted":false,"changed_token_positions":70,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":44,"collapsed_token_delta":-1,"collapsed_tuple_leak_window_count":44,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":9,"measurement_hash":"8a62589faf940a247b8807578eba3f20aa58bb856deda05dcdc7d1999a89b789","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":28,"root_window_count":60,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","cycle7-expand-its-like","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-colon","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","expand-cannot","surface-space-after-any-word","cycle7-expand-i-have","surface-space-after-comma"],"token_delta":10,"tuple_leak_window_count":9},"label":"watermarked","pristine_detected":true,"pristine_score":0.653030303030303,"sample_id":"cycle7-stage-a-structured_instructional-watermarked","sanitizers":{"cf_strip":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"nfkc":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"nfkc_cf_strip":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"raw":{"detected":false,"score":0.5063852813852814,"text_hash":"0c108eb3ddccf5ede230d9f40ce7571fa075f40e0dad35fa859ffdd586d5e136"},"ws_collapse":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"},"ws_collapse_nfkc_cf_strip":{"detected":true,"score":0.6004964903269988,"text_hash":"1ab04446d34d14c3f69fecbedea60ab905b421aa47d624f0f17c894512018dc9"}}}],"secret_access_used_for_selection":false,"seed_base":810000,"summaries":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5254488114855487,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6426760741438237},"cycle7_combined":{"arm_id":"cycle7_combined","median_raw_watermarked_score":0.5332779853327798,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":1,"raw_watermarked_mean_score":0.5311415879052865,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5606060606060607,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6278505912405954},"cycle7_durable":{"arm_id":"cycle7_durable","median_raw_watermarked_score":0.6248778103616813,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":4,"raw_watermarked_mean_score":0.6248844377704097,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.6527570789865872,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6248844377704097}},"threshold":0.5570987654320988} diff --git a/evidence/cycle7-stage-a-2026-08-25/family2-density.json b/evidence/cycle7-stage-a-2026-08-25/family2-density.json new file mode 100644 index 00000000..6f6b854f --- /dev/null +++ b/evidence/cycle7-stage-a-2026-08-25/family2-density.json @@ -0,0 +1 @@ +{"algorithm_version":"cycle7-stage-a-density-v1","artifact_hash":"c4b26941fad816f9fe150c80124679974d5079870c3fd1c28d3a23b07a6f610b","detector_access_used_for_selection":false,"durable_catalog_version":"cycle7-durable-rule-catalog-v2","rows":[{"apostrophe_candidate_count":0,"apostrophe_rule_ids":[],"candidate_count":3,"compound_candidate_count":3,"compound_rule_ids":["lexical-compound-hyphenate-proof-of-concept","lexical-compound-hyphenate-proof-of-concept","lexical-compound-hyphenate-proof-of-concept"],"contraction_candidate_count":0,"rule_ids":["lexical-compound-hyphenate-proof-of-concept","lexical-compound-hyphenate-proof-of-concept","lexical-compound-hyphenate-proof-of-concept"],"sample_id":"cycle7-stage-a-general_explanatory-unwatermarked"},{"apostrophe_candidate_count":2,"apostrophe_rule_ids":["lexical-apostrophe-ascii-to-typographic","lexical-apostrophe-ascii-to-typographic"],"candidate_count":3,"compound_candidate_count":0,"compound_rule_ids":[],"contraction_candidate_count":1,"rule_ids":["lexical-apostrophe-ascii-to-typographic","cycle7-expand-its-important","lexical-apostrophe-ascii-to-typographic"],"sample_id":"cycle7-stage-a-general_explanatory-watermarked"},{"apostrophe_candidate_count":1,"apostrophe_rule_ids":["lexical-apostrophe-ascii-to-typographic"],"candidate_count":2,"compound_candidate_count":0,"compound_rule_ids":[],"contraction_candidate_count":1,"rule_ids":["cycle7-expand-its-an","lexical-apostrophe-ascii-to-typographic"],"sample_id":"cycle7-stage-a-technical_explanation-unwatermarked"},{"apostrophe_candidate_count":0,"apostrophe_rule_ids":[],"candidate_count":1,"compound_candidate_count":1,"compound_rule_ids":["lexical-compound-open-proof-of-concept"],"contraction_candidate_count":0,"rule_ids":["lexical-compound-open-proof-of-concept"],"sample_id":"cycle7-stage-a-technical_explanation-watermarked"},{"apostrophe_candidate_count":3,"apostrophe_rule_ids":["lexical-apostrophe-ascii-to-typographic","lexical-apostrophe-ascii-to-typographic","lexical-apostrophe-ascii-to-typographic"],"candidate_count":7,"compound_candidate_count":0,"compound_rule_ids":[],"contraction_candidate_count":4,"rule_ids":["expand-you-are","lexical-apostrophe-ascii-to-typographic","expand-do-not","lexical-apostrophe-ascii-to-typographic","expand-that-is-not-x","expand-is-not","lexical-apostrophe-ascii-to-typographic"],"sample_id":"cycle7-stage-a-conversational_prose-unwatermarked"},{"apostrophe_candidate_count":0,"apostrophe_rule_ids":[],"candidate_count":1,"compound_candidate_count":1,"compound_rule_ids":["lexical-compound-hyphenate-point-of-view"],"contraction_candidate_count":0,"rule_ids":["lexical-compound-hyphenate-point-of-view"],"sample_id":"cycle7-stage-a-conversational_prose-watermarked"},{"apostrophe_candidate_count":1,"apostrophe_rule_ids":["lexical-apostrophe-ascii-to-typographic"],"candidate_count":1,"compound_candidate_count":0,"compound_rule_ids":[],"contraction_candidate_count":0,"rule_ids":["lexical-apostrophe-ascii-to-typographic"],"sample_id":"cycle7-stage-a-structured_instructional-unwatermarked"},{"apostrophe_candidate_count":3,"apostrophe_rule_ids":["lexical-apostrophe-ascii-to-typographic","lexical-apostrophe-ascii-to-typographic","lexical-apostrophe-ascii-to-typographic"],"candidate_count":6,"compound_candidate_count":0,"compound_rule_ids":[],"contraction_candidate_count":3,"rule_ids":["cycle7-expand-its-like","lexical-apostrophe-ascii-to-typographic","expand-cannot","lexical-apostrophe-ascii-to-typographic","cycle7-expand-i-have","lexical-apostrophe-ascii-to-typographic"],"sample_id":"cycle7-stage-a-structured_instructional-watermarked"}],"seed_base":810000} diff --git a/evidence/cycle7-stage-a-2026-08-25/fixture-geometry.json b/evidence/cycle7-stage-a-2026-08-25/fixture-geometry.json index b705cbf7..edf0e666 100644 --- a/evidence/cycle7-stage-a-2026-08-25/fixture-geometry.json +++ b/evidence/cycle7-stage-a-2026-08-25/fixture-geometry.json @@ -1 +1 @@ -{"algorithm_version":"cycle7-stage-a-fixture-compare-v1","arm_ids":["cycle6_spacing_b14","cycle7_durable","cycle7_combined"],"artifact_hash":"f76f56efcdad95c38fce42289cfa6a2689bf99436e37fd63d237db9b144c80ad","budget":14,"detector":{"available":true,"combined":{"arm_id":"cycle7_combined","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5239653475020847,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6295426209679976},"cycle6_spacing":{"arm_id":"cycle6_spacing_b14","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5254488114855487,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6426760741438237},"durable":{"arm_id":"cycle7_durable","median_raw_watermarked_score":0.6286616161616162,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":4,"raw_watermarked_mean_score":0.6275224189477956,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.6634137990070192,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6275224189477956}},"rows":[{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":77,"candidate_exhausted":false,"changed_token_positions":473,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":408,"collapsed_token_delta":-3,"collapsed_tuple_leak_window_count":411,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":356,"measurement_hash":"67c9eb7ce713eebb51900bbd991e28abdd07873491114baf405a410c0ad13fe6","protected_blocked_count":4,"quote_blocked_count":1,"reachable_unselected_static_cover":51,"root_window_count":426,"selected_count":14,"selected_rule_ids":["contract-are-not","contract-do-not","surface-space-after-any-word","contract-cannot","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-before-sentence","surface-space-before-sentence"],"token_delta":8,"tuple_leak_window_count":378},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":81,"candidate_exhausted":false,"changed_token_positions":418,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":396,"collapsed_token_delta":-7,"collapsed_tuple_leak_window_count":399,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":353,"measurement_hash":"8b9061d597e0f732a02c050de3620e93f2ec05acd9dcfc5d97dc1082cb7d3db7","protected_blocked_count":4,"quote_blocked_count":0,"reachable_unselected_static_cover":51,"root_window_count":426,"selected_count":14,"selected_rule_ids":["contract-are-not","contract-do-not","surface-space-after-any-word","contract-cannot","cycle7-contract-they-have","surface-space-after-any-word","cycle7-ortho-amongst-among","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-before-sentence","surface-space-before-sentence"],"token_delta":2,"tuple_leak_window_count":372},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":9,"candidate_exhausted":true,"changed_token_positions":475,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":381,"collapsed_token_delta":-10,"collapsed_tuple_leak_window_count":382,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":381,"measurement_hash":"675f24e72b0b7dd4ac7a0b5f3c25b55c8122e28e3dca8fff20e5a2a2abd999fe","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":426,"selected_count":8,"selected_rule_ids":["contract-are-not","contract-do-not","contract-cannot","contract-i-am","cycle7-contract-they-have","cycle7-ortho-towards-toward","cycle7-ortho-amongst-among","contract-do-not"],"token_delta":-10,"tuple_leak_window_count":382}},"source_sample_id":"contraction-rich","source_text_hash":"75445d3d08c371e274c0c6a1fb9b98e374e0eb60e5cf373a7921499771d41e91"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":18,"candidate_exhausted":false,"changed_token_positions":137,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":132,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":132,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":68,"measurement_hash":"50dbcb37bbc266eb6f5cccff5f597629f059f4730eb84c5abb545baf9aefe150","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":2,"root_window_count":132,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":94},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":18,"candidate_exhausted":false,"changed_token_positions":137,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":132,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":132,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":68,"measurement_hash":"e765da5b67fc654decea5cf2a1ebda2daa2addabd420210e3abdb3ee1d39166c","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":2,"root_window_count":132,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":94},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":132,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":132,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":132,"measurement_hash":"cb8747086fa20595a37904be0cb07aff1d90e82557a8b5086e28253e750a1460","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":132,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":132}},"source_sample_id":"contraction-sparse","source_text_hash":"ef1e423456e6e8bf14408dd5e80ea6a491938c17541dd00a3fa90695ebecbed3"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":false,"candidate_count":10,"candidate_exhausted":true,"changed_token_positions":61,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":51,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":51,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":13,"measurement_hash":"20550c7fb44d950468245109b89a29d7185a2b5703fae7fafeed03c169277809","protected_blocked_count":0,"quote_blocked_count":3,"reachable_unselected_static_cover":0,"root_window_count":51,"selected_count":10,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":10,"tuple_leak_window_count":23},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":false,"candidate_count":13,"candidate_exhausted":true,"changed_token_positions":56,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":-2,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":17,"measurement_hash":"f23745fd66429f0f64e1388eec798c49f8bb13b47bf2cd2d7c5bd9228efdbec8","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":51,"selected_count":8,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-comma","contract-they-are","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","contract-do-not"],"token_delta":4,"tuple_leak_window_count":24},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":3,"candidate_exhausted":true,"changed_token_positions":39,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":-2,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":40,"measurement_hash":"c2681e52e67c056c661046bc533cf823801f91c6b552e8265681af76aa26a31b","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":51,"selected_count":2,"selected_rule_ids":["contract-they-are","contract-do-not"],"token_delta":-2,"tuple_leak_window_count":40}},"source_sample_id":"quote-interior","source_text_hash":"9435a82cc500b4ec86c09b20965a484ebea69f0ae0603cdf0a057a9b632c93ba"}]} +{"algorithm_version":"cycle7-stage-a-fixture-compare-v1","arm_ids":["cycle6_spacing_b14","cycle7_durable","cycle7_combined"],"artifact_hash":"e5aab17260087baa64ff58c48016e382883fd8dee455c58c507036cc5d1af27b","budget":14,"detector":{"available":true,"combined":{"arm_id":"cycle7_combined","median_raw_watermarked_score":0.5332779853327798,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":1,"raw_watermarked_mean_score":0.5311415879052865,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5606060606060607,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6278505912405954},"cycle6_spacing":{"arm_id":"cycle6_spacing_b14","median_raw_watermarked_score":0.5242970242970243,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":0,"raw_watermarked_mean_score":0.5254488114855487,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.5473293541475359,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6426760741438237},"durable":{"arm_id":"cycle7_durable","median_raw_watermarked_score":0.6248778103616813,"pristine_unwatermarked_detected":0,"pristine_watermarked_detected":4,"raw_unwatermarked_detected":0,"raw_watermarked_detected":4,"raw_watermarked_mean_score":0.6248844377704097,"unwatermarked_row_count":4,"watermarked_row_count":4,"worst_raw_watermarked_score":0.6527570789865872,"ws_collapse_unwatermarked_detected":0,"ws_collapse_watermarked_detected":4,"ws_collapse_watermarked_mean_score":0.6248844377704097}},"rows":[{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":77,"candidate_exhausted":false,"changed_token_positions":473,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":408,"collapsed_token_delta":-3,"collapsed_tuple_leak_window_count":411,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":356,"measurement_hash":"67c9eb7ce713eebb51900bbd991e28abdd07873491114baf405a410c0ad13fe6","protected_blocked_count":4,"quote_blocked_count":1,"reachable_unselected_static_cover":51,"root_window_count":426,"selected_count":14,"selected_rule_ids":["contract-are-not","contract-do-not","surface-space-after-any-word","contract-cannot","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-before-sentence","surface-space-before-sentence"],"token_delta":8,"tuple_leak_window_count":378},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":81,"candidate_exhausted":false,"changed_token_positions":418,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":396,"collapsed_token_delta":-7,"collapsed_tuple_leak_window_count":399,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":353,"measurement_hash":"8b9061d597e0f732a02c050de3620e93f2ec05acd9dcfc5d97dc1082cb7d3db7","protected_blocked_count":4,"quote_blocked_count":0,"reachable_unselected_static_cover":51,"root_window_count":426,"selected_count":14,"selected_rule_ids":["contract-are-not","contract-do-not","surface-space-after-any-word","contract-cannot","cycle7-contract-they-have","surface-space-after-any-word","cycle7-ortho-amongst-among","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-before-sentence","surface-space-before-sentence"],"token_delta":2,"tuple_leak_window_count":372},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":9,"candidate_exhausted":true,"changed_token_positions":475,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":381,"collapsed_token_delta":-10,"collapsed_tuple_leak_window_count":382,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":381,"measurement_hash":"675f24e72b0b7dd4ac7a0b5f3c25b55c8122e28e3dca8fff20e5a2a2abd999fe","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":426,"selected_count":8,"selected_rule_ids":["contract-are-not","contract-do-not","contract-cannot","contract-i-am","cycle7-contract-they-have","cycle7-ortho-towards-toward","cycle7-ortho-amongst-among","contract-do-not"],"token_delta":-10,"tuple_leak_window_count":382}},"source_sample_id":"contraction-rich","source_text_hash":"75445d3d08c371e274c0c6a1fb9b98e374e0eb60e5cf373a7921499771d41e91"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":18,"candidate_exhausted":false,"changed_token_positions":137,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":132,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":132,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":68,"measurement_hash":"50dbcb37bbc266eb6f5cccff5f597629f059f4730eb84c5abb545baf9aefe150","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":2,"root_window_count":132,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":94},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":18,"candidate_exhausted":false,"changed_token_positions":137,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":132,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":132,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":68,"measurement_hash":"e765da5b67fc654decea5cf2a1ebda2daa2addabd420210e3abdb3ee1d39166c","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":2,"root_window_count":132,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-same","surface-space-after-any-word","surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":94},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":0,"candidate_exhausted":true,"changed_token_positions":0,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":132,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":132,"failure_classes":["insufficient_transform_density","candidate_exhaustion"],"intact_window_count":132,"measurement_hash":"cb8747086fa20595a37904be0cb07aff1d90e82557a8b5086e28253e750a1460","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":132,"selected_count":0,"selected_rule_ids":[],"token_delta":0,"tuple_leak_window_count":132}},"source_sample_id":"contraction-sparse","source_text_hash":"ef1e423456e6e8bf14408dd5e80ea6a491938c17541dd00a3fa90695ebecbed3"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":true,"candidate_count":26,"candidate_exhausted":false,"changed_token_positions":131,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":120,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":120,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":61,"measurement_hash":"5455f18112026f44867ba067a8f003c9a85baa3434ffadd14762102ba7e702c9","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":9,"root_window_count":120,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-any-word","surface-space-after-point","surface-space-after-any-word","surface-space-after-need","surface-space-after-any-word","surface-space-after-period","surface-space-before-sentence","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":14,"tuple_leak_window_count":81},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":true,"candidate_count":32,"candidate_exhausted":false,"changed_token_positions":131,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":70,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":73,"failure_classes":["budget_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":36,"measurement_hash":"28affa9b64ff6a1ddd0092a51012b3c3e91a22737537713607aa6f25d462b595","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":120,"selected_count":14,"selected_rule_ids":["surface-space-after-any-word","lexical-compound-hyphenate-proof-of-concept","surface-space-after-any-word","surface-space-after-any-word","lexical-compound-hyphenate-point-of-view","surface-space-after-need","lexical-compound-open-step-by-step","surface-space-after-period","lexical-compound-hyphenate-state-of-the-art","surface-space-after-any-word","surface-space-after-any-word","lexical-compound-hyphenate-face-to-face","surface-space-after-any-word","lexical-compound-open-case-by-case"],"token_delta":8,"tuple_leak_window_count":52},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":6,"candidate_exhausted":true,"changed_token_positions":13,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":70,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":73,"failure_classes":["candidate_exhaustion"],"intact_window_count":70,"measurement_hash":"e26aeb3c7ba0c9d1840fa966c0c14dec5656bc9ed961c2b8d6c0ee79f0f04859","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":120,"selected_count":6,"selected_rule_ids":["lexical-compound-hyphenate-proof-of-concept","lexical-compound-hyphenate-point-of-view","lexical-compound-open-step-by-step","lexical-compound-hyphenate-state-of-the-art","lexical-compound-hyphenate-face-to-face","lexical-compound-open-case-by-case"],"token_delta":0,"tuple_leak_window_count":73}},"source_sample_id":"compound-rich","source_text_hash":"833befced9a79a41d3512520c3661d40000a09a8126a96e7ceca550d2013860f"},{"arms":{"cycle6_spacing_b14":{"arm_id":"cycle6_spacing_b14","budget_exhausted":false,"candidate_count":10,"candidate_exhausted":true,"changed_token_positions":61,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":true,"collapsed_intact_window_count":51,"collapsed_token_delta":0,"collapsed_tuple_leak_window_count":51,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery","whitespace_only_channel"],"intact_window_count":13,"measurement_hash":"20550c7fb44d950468245109b89a29d7185a2b5703fae7fafeed03c169277809","protected_blocked_count":0,"quote_blocked_count":3,"reachable_unselected_static_cover":0,"root_window_count":51,"selected_count":10,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-comma","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word"],"token_delta":10,"tuple_leak_window_count":23},"cycle7_combined":{"arm_id":"cycle7_combined","budget_exhausted":false,"candidate_count":13,"candidate_exhausted":true,"changed_token_positions":56,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":-2,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":17,"measurement_hash":"f23745fd66429f0f64e1388eec798c49f8bb13b47bf2cd2d7c5bd9228efdbec8","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":51,"selected_count":8,"selected_rule_ids":["surface-space-after-any-word","surface-space-after-comma","contract-they-are","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","surface-space-after-any-word","contract-do-not"],"token_delta":4,"tuple_leak_window_count":24},"cycle7_durable":{"arm_id":"cycle7_durable","budget_exhausted":false,"candidate_count":3,"candidate_exhausted":true,"changed_token_positions":39,"closure_free":false,"collapsed_closure_free":false,"collapsed_equals_collapsed_source":false,"collapsed_intact_window_count":40,"collapsed_token_delta":-2,"collapsed_tuple_leak_window_count":40,"failure_classes":["candidate_exhaustion","tokenizer_resynchronization_or_long_range_recovery"],"intact_window_count":40,"measurement_hash":"c2681e52e67c056c661046bc533cf823801f91c6b552e8265681af76aa26a31b","protected_blocked_count":0,"quote_blocked_count":0,"reachable_unselected_static_cover":0,"root_window_count":51,"selected_count":2,"selected_rule_ids":["contract-they-are","contract-do-not"],"token_delta":-2,"tuple_leak_window_count":40}},"source_sample_id":"quote-interior","source_text_hash":"9435a82cc500b4ec86c09b20965a484ebea69f0ae0603cdf0a057a9b632c93ba"}]} diff --git a/fuckmark/cycle7/density.py b/fuckmark/cycle7/density.py new file mode 100644 index 00000000..6c541c84 --- /dev/null +++ b/fuckmark/cycle7/density.py @@ -0,0 +1,38 @@ +from __future__ import annotations + +from collections.abc import Mapping, Sequence + +from .registry import cycle7_durable_transform_registry + + +def durable_candidate_rule_ids(text: str) -> tuple[str, ...]: + enumeration = cycle7_durable_transform_registry().enumerate(text) + return tuple(candidate.rule_id for candidate in enumeration.candidates) + + +def durable_density_row(sample_id: str, text: str) -> dict[str, object]: + rule_ids = durable_candidate_rule_ids(text) + compound_ids = tuple(rule_id for rule_id in rule_ids if rule_id.startswith("lexical-compound-")) + contraction_ids = tuple( + rule_id + for rule_id in rule_ids + if rule_id.startswith(("contract-", "expand-", "cycle7-contract-", "cycle7-expand-")) + ) + apostrophe_ids = tuple(rule_id for rule_id in rule_ids if rule_id.startswith("lexical-apostrophe-")) + return { + "sample_id": sample_id, + "candidate_count": len(rule_ids), + "compound_candidate_count": len(compound_ids), + "contraction_candidate_count": len(contraction_ids), + "apostrophe_candidate_count": len(apostrophe_ids), + "rule_ids": rule_ids, + "compound_rule_ids": compound_ids, + "apostrophe_rule_ids": apostrophe_ids, + } + + +def durable_density_table(samples: Sequence[Mapping[str, object]]) -> tuple[dict[str, object], ...]: + rows: list[dict[str, object]] = [] + for sample in samples: + rows.append(durable_density_row(str(sample["sample_id"]), str(sample["text"]))) + return tuple(rows) diff --git a/fuckmark/cycle7/durable_rules.py b/fuckmark/cycle7/durable_rules.py index 5a780042..529416a4 100644 --- a/fuckmark/cycle7/durable_rules.py +++ b/fuckmark/cycle7/durable_rules.py @@ -10,13 +10,13 @@ zrd_forward_contraction_extension_rules, zrd_reverse_contraction_extension_rules, ) -from ..transforms.lexical_rules import development_lexical_rules +from ..transforms.lexical_rules import LexicalConstruction, LexicalTemplateRule, development_lexical_rules from ..transforms.rules import LiteralTransformRule from ..transforms.schema import TransformFamily, TransformTier from ..transforms.syntax_rules import development_syntax_rules -CYCLE7_DURABLE_RULE_CATALOG_VERSION = "cycle7-durable-rule-catalog-v1" +CYCLE7_DURABLE_RULE_CATALOG_VERSION = "cycle7-durable-rule-catalog-v2" @dataclass(frozen=True, slots=True) @@ -310,6 +310,61 @@ def cycle7_orthography_rules() -> tuple[LiteralTransformRule, ...]: return _rules_from_metadata(cycle7_orthography_metadata()) +CYCLE7_ATTESTED_COMPOUND_PAIRS = ( + ("proof of concept", "proof-of-concept"), + ("point of view", "point-of-view"), + ("step by step", "step-by-step"), + ("case by case", "case-by-case"), + ("end to end", "end-to-end"), + ("state of the art", "state-of-the-art"), + ("face to face", "face-to-face"), +) + + +def cycle7_compound_rules() -> tuple[LexicalTemplateRule, ...]: + rules: list[LexicalTemplateRule] = [] + for open_form, hyphen_form in CYCLE7_ATTESTED_COMPOUND_PAIRS: + slug = hyphen_form + rules.append( + LexicalTemplateRule.create( + rule_id=f"lexical-compound-hyphenate-{slug}", + version="v1", + source=open_form, + replacement=hyphen_form, + construction=LexicalConstruction.ATTESTED_OPEN_HYPHEN_COMPOUND, + ) + ) + rules.append( + LexicalTemplateRule.create( + rule_id=f"lexical-compound-open-{slug}", + version="v1", + source=hyphen_form, + replacement=open_form, + construction=LexicalConstruction.ATTESTED_OPEN_HYPHEN_COMPOUND, + ) + ) + return tuple(rules) + + +def cycle7_typographic_apostrophe_rules() -> tuple[LexicalTemplateRule, ...]: + return ( + LexicalTemplateRule.create( + rule_id="lexical-apostrophe-ascii-to-typographic", + version="v1", + source="'", + replacement="\u2019", + construction=LexicalConstruction.INWORD_TYPOGRAPHIC_APOSTROPHE, + ), + LexicalTemplateRule.create( + rule_id="lexical-apostrophe-typographic-to-ascii", + version="v1", + source="\u2019", + replacement="'", + construction=LexicalConstruction.INWORD_TYPOGRAPHIC_APOSTROPHE, + ), + ) + + def cycle7_durable_rules() -> tuple[object, ...]: return ( *development_forward_contraction_rules(), @@ -319,6 +374,8 @@ def cycle7_durable_rules() -> tuple[object, ...]: *cycle7_new_contraction_rules(), *cycle7_bounded_copula_rules(), *cycle7_orthography_rules(), + *cycle7_compound_rules(), + *cycle7_typographic_apostrophe_rules(), *development_lexical_rules(), *development_syntax_rules(), ) diff --git a/fuckmark/cycle7/fixtures.py b/fuckmark/cycle7/fixtures.py index dd1e51b5..064afe75 100644 --- a/fuckmark/cycle7/fixtures.py +++ b/fuckmark/cycle7/fixtures.py @@ -20,6 +20,11 @@ "same experimental apparatus. Calibration uses a held-out population." ) +COMPOUND_RICH = ( + "A proof of concept and a point of view need step-by-step checks. " + "State of the art methods stay face to face with case-by-case review." +) + QUOTE_INTERIOR = 'He answered, "They are not finished and we do not agree."' AMBIGUOUS_NEGATIVES = ( @@ -46,6 +51,7 @@ def fixture_samples() -> tuple[tuple[str, str], ...]: return ( ("contraction-rich", CONTRACTION_RICH), ("contraction-sparse", CONTRACTION_SPARSE), + ("compound-rich", COMPOUND_RICH), ("quote-interior", QUOTE_INTERIOR), ) diff --git a/fuckmark/cycle7/registry.py b/fuckmark/cycle7/registry.py index f4a918d0..f72f47ce 100644 --- a/fuckmark/cycle7/registry.py +++ b/fuckmark/cycle7/registry.py @@ -11,7 +11,7 @@ from .durable_rules import cycle7_durable_rules -CYCLE7_DURABLE_REGISTRY_ID = "cycle7-durable-contraction-orthography-v1" +CYCLE7_DURABLE_REGISTRY_ID = "cycle7-durable-catalog-v2" CYCLE7_COMBINED_REGISTRY_ID = "cycle7-durable-plus-cycle6-spacing-v1" diff --git a/fuckmark/transforms/cycle7_quote_policy.py b/fuckmark/transforms/cycle7_quote_policy.py index 8744f7e1..7ed00755 100644 --- a/fuckmark/transforms/cycle7_quote_policy.py +++ b/fuckmark/transforms/cycle7_quote_policy.py @@ -40,6 +40,8 @@ "cycle7-contract-", "cycle7-expand-", "cycle7-ortho-", + "lexical-compound-", + "lexical-apostrophe-", "lexical-", "syntax-", ) diff --git a/fuckmark/transforms/lexical_rules.py b/fuckmark/transforms/lexical_rules.py index 852f8cb4..0f9b2b24 100644 --- a/fuckmark/transforms/lexical_rules.py +++ b/fuckmark/transforms/lexical_rules.py @@ -14,6 +14,8 @@ class LexicalConstruction(str, Enum): SENTENCE_INITIAL_DISCOURSE_MARKER = "sentence_initial_discourse_marker" + ATTESTED_OPEN_HYPHEN_COMPOUND = "attested_open_hyphen_compound" + INWORD_TYPOGRAPHIC_APOSTROPHE = "inword_typographic_apostrophe" @dataclass(frozen=True, slots=True) @@ -139,19 +141,53 @@ def precondition(self, text: str, start: int, end: int) -> bool: raise ValueError("lexical precondition span is outside text") if text[start:end].casefold() != self.source.casefold(): raise ValueError("lexical precondition span does not match rule source") - if end >= len(text) or not text[end].isspace(): - return False if self.construction is LexicalConstruction.SENTENCE_INITIAL_DISCOURSE_MARKER: + if end >= len(text) or not text[end].isspace(): + return False prefix = text[:start] stripped = prefix.rstrip() if stripped and stripped[-1] not in ".?!": return False + elif self.construction is LexicalConstruction.ATTESTED_OPEN_HYPHEN_COMPOUND: + if not attested_open_hyphen_compound_span_admissible(text, start, end): + return False + elif self.construction is LexicalConstruction.INWORD_TYPOGRAPHIC_APOSTROPHE: + if not inword_typographic_apostrophe_span_admissible(text, start, end): + return False + else: + return False context = text[max(0, start - 96) : min(len(text), end + 96)].casefold() if any(value.casefold() in context for value in self.ambiguity_blacklist): return False return True +def attested_open_hyphen_compound_span_admissible(text: str, start: int, end: int) -> bool: + if start > 0: + previous = text[start - 1] + if previous.isalnum() or previous in "_-/\\": + return False + if end < len(text): + nxt = text[end] + if nxt.isalnum() or nxt in "_-/\\": + return False + return "--" not in text[start:end] + + +def inword_typographic_apostrophe_span_admissible(text: str, start: int, end: int) -> bool: + if start == 0 or end >= len(text): + return False + if end - start != 1: + return False + if text[start] not in {"'", "\u2019"}: + return False + if not text[start - 1].isalpha() or not text[end].isalpha(): + return False + if text[start - 1].isupper() and text[end].isupper(): + return False + return True + + def development_lexical_rules() -> tuple[LexicalTemplateRule, ...]: return ( LexicalTemplateRule.create( diff --git a/tests/test_cycle7_apostrophe.py b/tests/test_cycle7_apostrophe.py new file mode 100644 index 00000000..02343995 --- /dev/null +++ b/tests/test_cycle7_apostrophe.py @@ -0,0 +1,81 @@ +from fuckmark.cycle7.density import durable_density_row +from fuckmark.cycle7.durable_rules import cycle7_typographic_apostrophe_rules +from fuckmark.cycle7.registry import cycle7_durable_transform_registry +from fuckmark.cycle7.whitespace_collapse import collapse_horizontal_ascii_whitespace, sanitize_cycle7_variant +from fuckmark.transforms.hard_invariants import validate_hard_invariants +from fuckmark.transforms.lexical_rules import LexicalConstruction +from fuckmark.transforms.schema import CandidateRejectionReason, InvariantStatus, TransformFamily + + +def test_typographic_apostrophe_rules_are_inword_and_reversible() -> None: + rules = cycle7_typographic_apostrophe_rules() + assert len(rules) == 2 + assert all(rule.construction is LexicalConstruction.INWORD_TYPOGRAPHIC_APOSTROPHE for rule in rules) + assert all(rule.family is TransformFamily.LEXICAL_TEMPLATE for rule in rules) + registry = cycle7_durable_transform_registry() + text = "It's ready and we don't wait." + enumeration = registry.enumerate(text) + selected = tuple( + candidate.candidate_id + for candidate in enumeration.candidates + if candidate.rule_id == "lexical-apostrophe-ascii-to-typographic" + ) + assert len(selected) == 2 + result = registry.apply(enumeration, selected) + assert "'" not in result.output_text + assert "\u2019" in result.output_text + assert collapse_horizontal_ascii_whitespace(result.output_text) == result.output_text + for variant in ( + "raw", + "nfkc", + "cf_strip", + "nfkc_cf_strip", + "ws_collapse", + "ws_collapse_nfkc_cf_strip", + ): + assert sanitize_cycle7_variant(variant, result.output_text) == result.output_text + inverse = registry.enumerate(result.output_text) + reverse = tuple( + candidate.candidate_id + for candidate in inverse.candidates + if candidate.rule_id == "lexical-apostrophe-typographic-to-ascii" + ) + restored = registry.apply(inverse, reverse) + assert restored.output_text == text + report = validate_hard_invariants(text, result.output_text) + assert report.status is InvariantStatus.PASS + + +def test_typographic_apostrophe_skips_quote_delimiters_and_plural_possessives() -> None: + registry = cycle7_durable_transform_registry() + quoted = registry.enumerate("'Keep this quoted.'") + assert not any( + candidate.rule_id.startswith("lexical-apostrophe-") for candidate in quoted.candidates + ) + possessive = registry.enumerate("The students' notes remained.") + assert not any( + candidate.rule_id.startswith("lexical-apostrophe-") for candidate in possessive.candidates + ) + + +def test_typographic_apostrophe_all_caps_and_code_are_blocked() -> None: + registry = cycle7_durable_transform_registry() + all_caps = registry.enumerate("WE DON'T STOP.") + assert not any( + candidate.rule_id.startswith("lexical-apostrophe-") for candidate in all_caps.candidates + ) + assert any( + rejection.rule_id == "lexical-apostrophe-ascii-to-typographic" + and rejection.reason is CandidateRejectionReason.PRECONDITION_FAILED + for rejection in all_caps.rejections + ) + code = registry.enumerate("Run `don't` as a literal command token.") + assert not any( + candidate.rule_id.startswith("lexical-apostrophe-") for candidate in code.candidates + ) + + +def test_typographic_apostrophe_density_on_contraction_prose() -> None: + row = durable_density_row("apostrophe-prose", "It's not uncommon, and I've seen it.") + assert "lexical-apostrophe-ascii-to-typographic" in row["rule_ids"] + assert int(row["candidate_count"]) >= 2 diff --git a/tests/test_cycle7_compounds.py b/tests/test_cycle7_compounds.py new file mode 100644 index 00000000..23b956f3 --- /dev/null +++ b/tests/test_cycle7_compounds.py @@ -0,0 +1,124 @@ +from fuckmark.cycle7.density import durable_density_row +from fuckmark.cycle7.durable_rules import cycle7_compound_rules +from fuckmark.cycle7.fixtures import COMPOUND_RICH +from fuckmark.cycle7.registry import cycle7_durable_transform_registry +from fuckmark.cycle7.whitespace_collapse import collapse_horizontal_ascii_whitespace +from fuckmark.transforms import ProtectedSpanExtractor +from fuckmark.transforms.lexical_rules import LexicalConstruction +from fuckmark.transforms.schema import CandidateRejectionReason, TransformFamily + + +def test_compound_rules_use_attested_construction() -> None: + rules = cycle7_compound_rules() + assert rules + assert all(rule.construction is LexicalConstruction.ATTESTED_OPEN_HYPHEN_COMPOUND for rule in rules) + assert all(rule.family is TransformFamily.LEXICAL_TEMPLATE for rule in rules) + assert all(" " in rule.source or "-" in rule.source for rule in rules) + assert all(" " not in rule.replacement for rule in rules) + + +def test_open_and_hyphenated_compounds_are_reversible_and_survive_collapse() -> None: + registry = cycle7_durable_transform_registry() + text = "A proof of concept needs a point-of-view check." + enumeration = registry.enumerate(text) + selected = tuple( + candidate.candidate_id + for candidate in enumeration.candidates + if candidate.rule_id + in { + "lexical-compound-hyphenate-proof-of-concept", + "lexical-compound-open-point-of-view", + } + ) + assert len(selected) == 2 + result = registry.apply(enumeration, selected) + assert result.output_text == "A proof-of-concept needs a point of view check." + assert collapse_horizontal_ascii_whitespace(result.output_text) == result.output_text + inverse = registry.enumerate(result.output_text) + reverse = tuple( + candidate.candidate_id + for candidate in inverse.candidates + if candidate.rule_id + in { + "lexical-compound-open-proof-of-concept", + "lexical-compound-hyphenate-point-of-view", + } + ) + restored = registry.apply(inverse, reverse) + assert restored.output_text == text + + +def test_compound_accepts_sentence_final_punctuation() -> None: + registry = cycle7_durable_transform_registry() + enumeration = registry.enumerate("This is a proof of concept.") + selected = tuple( + candidate + for candidate in enumeration.candidates + if candidate.rule_id == "lexical-compound-hyphenate-proof-of-concept" + ) + assert len(selected) == 1 + result = registry.apply(enumeration, (selected[0].candidate_id,)) + assert result.output_text == "This is a proof-of-concept." + + +def test_hyphen_chain_and_path_false_positives_are_blocked() -> None: + registry = cycle7_durable_transform_registry() + chained = registry.enumerate("See the proof-of-concept-note before shipping.") + assert not any( + candidate.rule_id == "lexical-compound-open-proof-of-concept" + for candidate in chained.candidates + ) + assert any( + rejection.rule_id == "lexical-compound-open-proof-of-concept" + and rejection.reason is CandidateRejectionReason.PRECONDITION_FAILED + for rejection in chained.rejections + ) + incomplete = registry.enumerate("The proof of conceptual coverage is separate.") + assert not any( + candidate.rule_id.startswith("lexical-compound-") for candidate in incomplete.candidates + ) + one_to_one_range = registry.enumerate("Count from one to one hundred carefully.") + assert not any( + "one-to-one" in candidate.rule_id for candidate in one_to_one_range.candidates + ) + + +def test_compound_rules_respect_protected_urls_dates_numbers_and_code() -> None: + registry = cycle7_durable_transform_registry() + cases = ( + "Visit https://example.com/proof-of-concept and continue.", + "On 2024-08-25 the proof of concept remains dated.", + "Run `proof of concept` as a literal command token.", + "Figure 3.5 reports the proof of concept result.", + ) + for text in cases: + enumeration = registry.enumerate(text) + protected = ProtectedSpanExtractor().extract(text).spans + for candidate in enumeration.candidates: + if not candidate.rule_id.startswith("lexical-compound-"): + continue + assert all( + not (candidate.start < span.end and span.start < candidate.end) + for span in protected + ), (text, candidate.rule_id, candidate.source_text) + + +def test_compound_density_on_compound_rich_fixture() -> None: + row = durable_density_row("compound-rich", COMPOUND_RICH) + assert int(row["compound_candidate_count"]) >= 4 + assert collapse_horizontal_ascii_whitespace(COMPOUND_RICH) == COMPOUND_RICH + + +def test_compound_inside_quotes_does_not_change_delimiters() -> None: + registry = cycle7_durable_transform_registry() + text = 'He said, "This proof of concept works."' + enumeration = registry.enumerate(text) + selected = tuple( + candidate + for candidate in enumeration.candidates + if candidate.rule_id == "lexical-compound-hyphenate-proof-of-concept" + ) + assert len(selected) == 1 + result = registry.apply(enumeration, (selected[0].candidate_id,)) + assert result.output_text == 'He said, "This proof-of-concept works."' + assert result.output_text.count('"') == text.count('"') diff --git a/tests/test_cycle7_durable_rules.py b/tests/test_cycle7_durable_rules.py index a8c83110..0dca5419 100644 --- a/tests/test_cycle7_durable_rules.py +++ b/tests/test_cycle7_durable_rules.py @@ -1,5 +1,6 @@ from fuckmark.cycle7.durable_rules import ( cycle7_bounded_copula_rules, + cycle7_compound_rules, cycle7_new_contraction_rules, cycle7_orthography_rules, rejected_ambiguous_contraction_examples, @@ -133,7 +134,11 @@ def test_punctuation_url_and_number_boundaries_do_not_fire_inside_tokens() -> No def test_durable_rules_are_not_spacing_family() -> None: - for rule in (*cycle7_new_contraction_rules(), *cycle7_orthography_rules()): - assert rule.family in {TransformFamily.CONTRACTION, TransformFamily.ORTHOGRAPHY} + for rule in (*cycle7_new_contraction_rules(), *cycle7_orthography_rules(), *cycle7_compound_rules()): + assert rule.family in { + TransformFamily.CONTRACTION, + TransformFamily.ORTHOGRAPHY, + TransformFamily.LEXICAL_TEMPLATE, + } assert not rule.source.endswith(" ") assert " " not in rule.replacement