Publish frozen TrellisMark blind-rephrasing corpus v1.0.0

.gitattributes

@@ -58,3 +58,4 @@ saved_model/**/* filter=lfs diff=lfs merge=lfs -text
 # Video files - compressed
 *.mp4 filter=lfs diff=lfs merge=lfs -text
 *.webm filter=lfs diff=lfs merge=lfs -text
+results/rephrase-residual-carry-v1.csv filter=lfs diff=lfs merge=lfs -text

README.md

@@ -0,0 +1,189 @@
+---
+pretty_name: TrellisMark Qwen3-4B Blind Rephrasing Corpus
+license: other
+language:
+- en
+task_categories:
+- text-generation
+size_categories:
+- 10K<n<100K
+tags:
+- watermarking
+- synthetic-data
+- ai-generated-text
+- paraphrase
+configs:
+- config_name: sources
+  data_files:
+  - split: corpus
+    path: sources/*.parquet
+- config_name: rewrites
+  default: true
+  data_files:
+  - split: corpus
+    path: rewrites/*.parquet
+- config_name: evidence
+  data_files:
+  - split: corpus
+    path: evidence/*.parquet
+---
+
+# TrellisMark Qwen3-4B blind rephrasing corpus
+
+This separate release contains the frozen detector-blind rephrasing experiment
+for **[TrellisMark](https://github.com/xlr8harder/trellismark)**. Its 16,384
+source documents are an exact document-ID-preserving subset of the
+**[main TrellisMark Qwen3-4B confirmation corpus](https://huggingface.co/datasets/xlr8harder/trellismark-qwen3-4b)**.
+It publishes both rewriters' outcome records, retained rewrite text and token
+IDs, aligned key-only and model-assisted evidence, and the reports behind the
+article's countermeasure figures.
+
+The rewriters received only a 200-Qwen-token source projection. They did not
+receive the watermark key, detector, user address, source prompt, or original
+preceding conversation. This release is separate from the main corpus because
+blind rewriting is a countermeasure dataset with its own models, acceptance
+contract, and failure modes.
+
+TrellisMark demonstrates a surveillance capability. This release exists so the
+claim and a practical countermeasure can be checked; it is not a recommendation
+to deploy user-level watermarking.
+
+## At a glance
+
+| Layer | Rows | What it contains |
+|---|---:|---|
+| `sources` | 16,384 | One 200-token marked source response for each rephrase job |
+| `rewrites` | 32,768 | One frozen outcome for every source under each of two blind rewriters |
+| `evidence` | 28,392 | Accepted rewrites with aligned key-only and model-assisted 32 x 4 soft evidence |
+
+| Rewriter | Outcomes | Length-valid rewrites | No length-valid response within budget |
+|---|---:|---:|---:|
+| Qwen3-4B | 16,384 | 13,531 | 2,853 |
+| Gemma 4 12B | 16,384 | 14,861 | 1,523 |
+
+The source pool contains 128 uniformly sampled held-out user addresses with 128
+independently prompted responses per address. Canonical source document IDs are
+unchanged from main-dataset revision `v1.0.2`, commit
+`c3ed277f6cb167bf8339e404828dcdc9099b04c4`.
+
+## Frozen rewrite contract
+
+- Source input: exactly the first 200 Qwen tokens of a marked response.
+- Rewriters: `Qwen/Qwen3-4B-Instruct-2507` at revision
+  `cdbee75f17c01a7cc42f958dc650907174af0554`, and
+  `google/gemma-4-12B-it` at revision
+  `707f0a3b8a3c7ad586ed01e27eafbad8a27dd0f7`.
+- Blindness: no key, detector, address, source prompt, or request history.
+- Acceptance: the first response with at least 200 Qwen tokens and a Qwen-token
+  length between 90% and 110% of the source length.
+- Content attempts: bounded; rows without a length-valid response retain an
+  explicit failure reason.
+- Analysis prefix: the first 200 Qwen tokens of each accepted rewrite.
+
+Length matching controls the amount of text available to the detector. It is
+not a semantic-quality guarantee. The residual-carry analysis separately
+reports multilingual embedding similarity and exact word/Qwen-token five-gram
+overlap, including detector-blind quality screens.
+
+## Load the data
+
+```python
+from datasets import load_dataset
+
+sources = load_dataset(
+    "xlr8harder/trellismark-qwen3-4b-rephrasing",
+    "sources",
+    split="corpus",
+)
+rewrites = load_dataset(
+    "xlr8harder/trellismark-qwen3-4b-rephrasing",
+    "rewrites",
+    split="corpus",
+)
+evidence = load_dataset(
+    "xlr8harder/trellismark-qwen3-4b-rephrasing",
+    "evidence",
+    split="corpus",
+)
+```
+
+`source_document_id` joins this release to the main corpus.
+`rewrite_document_id` uniquely identifies a rewriter/job pair. Accepted rows
+include token IDs in both the rewriter's native tokenizer and the Qwen tokenizer
+used by the detector. Rejected rows retain the frozen outcome and failure
+metadata but do not invent a retained rewrite.
+
+Each evidence row contains flattened 32 x 4 matrices under
+`key_only_position_by_symbol_scores` and
+`model_assisted_position_by_symbol_scores`. Reshape either list to `(32, 4)`
+before aggregation and exact Viterbi decoding. The original compressed NPZ
+copies used by the analysis are under `evidence_npz/`.
+
+## Reproduce the article results
+
+The `results/` directory contains:
+
+- `rephrase-curve-qwen3-4b-v1.json` and
+  `rephrase-curve-gemma4-12b-v1.json`: exact full-registry top-1 and top-10
+  recovery as independent accepted rewrites accumulate;
+- `rephrase-residual-carry-v1.csv`: row-level embedding similarity, word and
+  Qwen-token five-gram carry, token lengths, language labels, and attribution
+  diagnostics;
+- `rephrase-residual-carry-v1.json`: frozen thresholds, bootstrap summaries,
+  detector-blind screens, correlations, and interpretation.
+
+The standalone GitHub repository supplies the encoder and exact Viterbi
+decoder. Reproducing the published curves from cached evidence does not require
+either rewriter or the Qwen model. Rescoring new text with the model-assisted
+method does require the frozen Qwen model and scorer configuration.
+
+`export-manifest.json` records a SHA-256 digest for every release artifact and
+the source evidence commitments. The source jobs, public rows, accepted rows,
+and cached evidence sets are checked for one-to-one alignment during export.
+
+## Interpreting the quality screens
+
+Embedding similarity is used only as a coarse screen for obvious semantic
+drift. It is not a factuality judge. Token five-gram carry is the fraction of a
+rewrite's Qwen-token five-grams that occur anywhere in its frozen source
+prefix. It is mechanistically relevant because an unchanged five-token span
+preserves both an observed token and the four preceding tokens used for local
+watermark synchronization.
+
+The article comparison drops the bottom 10% of embedding similarity, the top
+10% of token-five-gram carry, or both, using quality/overlap variables alone.
+It does not inspect watermark evidence when deciding which rewrites to retain.
+These are descriptive screens rather than a causal decomposition of what makes
+watermark evidence survive.
+
+## Scope and limitations
+
+- This is an experimental existence proof and countermeasure study, not a
+  production security system.
+- Both attacks rewrite marked Qwen3-4B text; other source and rewriter models
+  may behave differently.
+- The length-only acceptance rule is intentionally simple and does not ensure
+  factual or stylistic equivalence.
+- Failed content attempts are represented by their frozen failure outcome; the
+  release does not reconstruct or promote rejected attempts as accepted text.
+- Reported recoveries are finite held-out observations, not population
+  guarantees.
+
+## Sources, attribution, and licensing
+
+Source and same-family rewrite text use
+[`Qwen/Qwen3-4B-Instruct-2507`](https://huggingface.co/Qwen/Qwen3-4B-Instruct-2507),
+whose model card lists Apache-2.0. Cross-family rewrites use
+[`google/gemma-4-12B-it`](https://huggingface.co/google/gemma-4-12B-it) under its
+model terms. Source prompt identifiers derive from
+[`allenai/WildChat`](https://huggingface.co/datasets/allenai/WildChat), whose
+card lists ODC-By; prompt text is not included here.
+
+The Hugging Face `license: other` label reflects the mixed contents of the
+repository. **Our original TrellisMark contribution is MIT licensed**,
+including the metadata schema, numeric evidence, analysis tables, and result
+reports. Generated prose and source-derived identifiers may remain subject to
+applicable upstream terms.
+
+TrellisMark is an independent research project. It is not affiliated with or
+endorsed by Google, Qwen, AllenAI, Hugging Face, or the Gemma team.

evidence/evidence-00000-of-00004.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0df59ab04f27adcb10473680f3c305c7751ae182d513e674b908fd537d26bea6
+size 11244923

evidence/evidence-00001-of-00004.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:77c6c5e809f9dc749a7516612132e7581315b37b28ffc9cac937d4f97b793176
+size 11243548

evidence/evidence-00002-of-00004.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:082dfd1cd8f4be3d6c17f7f8c3eaa75a8f0c06560aad31b1e184fe28f89996a2
+size 11242879

evidence/evidence-00003-of-00004.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7a91d0f1b28794f95150209e098cea79f2308a630b876d481ec974459b460937
+size 6432601

evidence_npz/gemma4_12b-evidence-200.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:03ea0cbad3da16a6f121b0e711158e139a43075e6b114002345d490907223701
+size 22530697

evidence_npz/qwen3_4b-evidence-200.npz

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b61bb37cdb4adc56692137f6a3336fa0f6cdc475d0fe2b98f86d5e106d6961bb
+size 20511976

export-manifest.json

@@ -0,0 +1,133 @@
+{
+  "accepted_rewrites": 28392,
+  "created_at": "2026-08-23T10:19:13.121848+00:00",
+  "files": [
+    {
+      "bytes": 11244923,
+      "path": "evidence/evidence-00000-of-00004.parquet",
+      "sha256": "0df59ab04f27adcb10473680f3c305c7751ae182d513e674b908fd537d26bea6"
+    },
+    {
+      "bytes": 11243548,
+      "path": "evidence/evidence-00001-of-00004.parquet",
+      "sha256": "77c6c5e809f9dc749a7516612132e7581315b37b28ffc9cac937d4f97b793176"
+    },
+    {
+      "bytes": 11242879,
+      "path": "evidence/evidence-00002-of-00004.parquet",
+      "sha256": "082dfd1cd8f4be3d6c17f7f8c3eaa75a8f0c06560aad31b1e184fe28f89996a2"
+    },
+    {
+      "bytes": 6432601,
+      "path": "evidence/evidence-00003-of-00004.parquet",
+      "sha256": "7a91d0f1b28794f95150209e098cea79f2308a630b876d481ec974459b460937"
+    },
+    {
+      "bytes": 22530697,
+      "path": "evidence_npz/gemma4_12b-evidence-200.npz",
+      "sha256": "03ea0cbad3da16a6f121b0e711158e139a43075e6b114002345d490907223701"
+    },
+    {
+      "bytes": 20511976,
+      "path": "evidence_npz/qwen3_4b-evidence-200.npz",
+      "sha256": "b61bb37cdb4adc56692137f6a3336fa0f6cdc475d0fe2b98f86d5e106d6961bb"
+    },
+    {
+      "bytes": 172687,
+      "path": "results/rephrase-curve-gemma4-12b-v1.json",
+      "sha256": "d28405f4e2c0c8d0495207efd1ba193141b1276f259ce396f889b177f1354aa3"
+    },
+    {
+      "bytes": 173877,
+      "path": "results/rephrase-curve-qwen3-4b-v1.json",
+      "sha256": "8123d68ab5fc4c34b3c45a3f61b57f57220d940ac242b13e25df19f28a8083cd"
+    },
+    {
+      "bytes": 11635385,
+      "path": "results/rephrase-residual-carry-v1.csv",
+      "sha256": "f4e5d94dbc8ed356fb2ba8509514b97db2f2557b417b6ced293ba1bd70f970cb"
+    },
+    {
+      "bytes": 150129,
+      "path": "results/rephrase-residual-carry-v1.json",
+      "sha256": "0b1cd0a37cd13982b7dc3b342be3b8814ea01cfacec35def6efa5a0e946c41d0"
+    },
+    {
+      "bytes": 7659197,
+      "path": "rewrites/rewrites-00000-of-00005.parquet",
+      "sha256": "e6af0d57988c1d4af493724e30c02bdadb30620accd8cd75a526ee1d14cf4e1b"
+    },
+    {
+      "bytes": 7711152,
+      "path": "rewrites/rewrites-00001-of-00005.parquet",
+      "sha256": "2ead5fb369b85bc7bfd27760f6e114450cdeb8ba498cf911f48af0150bc5517a"
+    },
+    {
+      "bytes": 7176675,
+      "path": "rewrites/rewrites-00002-of-00005.parquet",
+      "sha256": "7ccda18aa5bf0651182e392e48e25782b682ab24edc31f4d99859f9148dfeb51"
+    },
+    {
+      "bytes": 7208343,
+      "path": "rewrites/rewrites-00003-of-00005.parquet",
+      "sha256": "d1a1ca90d42de0f5ca13d150be417bf29a701a57cd9ad5b1c19f9d16a7c2ad7c"
+    },
+    {
+      "bytes": 807273,
+      "path": "rewrites/rewrites-00004-of-00005.parquet",
+      "sha256": "9ad8a66613edaac4c5b0b7a74bc2113554ae8adf0573f38a1041ad5a11ae34b4"
+    },
+    {
+      "bytes": 5593052,
+      "path": "sources/sources-00000-of-00003.parquet",
+      "sha256": "4bd6471f0f891fdec936188605d2be91c2d7b5d4cce3b941df8d88866448fd07"
+    },
+    {
+      "bytes": 5595457,
+      "path": "sources/sources-00001-of-00003.parquet",
+      "sha256": "308f3ee35a50ba330c119223765315d7684b5f4207126772dd67eefdff417e92"
+    },
+    {
+      "bytes": 317159,
+      "path": "sources/sources-00002-of-00003.parquet",
+      "sha256": "067318ffc46ca0a1f949caa74574616c345566955ab237ad2e8324bd6babdf05"
+    }
+  ],
+  "jobs": 16384,
+  "rewriters": {
+    "gemma4_12b": {
+      "accepted": 14861,
+      "failure_reasons": {
+        "no length-valid response within content-attempt budget": 1523
+      },
+      "model": "google/gemma-4-12B-it",
+      "model_revision": "707f0a3b8a3c7ad586ed01e27eafbad8a27dd0f7",
+      "outcomes": 16384,
+      "rejected": 1523,
+      "source_evidence_sha256": "03ea0cbad3da16a6f121b0e711158e139a43075e6b114002345d490907223701"
+    },
+    "qwen3_4b": {
+      "accepted": 13531,
+      "failure_reasons": {
+        "no length-valid response within content-attempt budget": 2853
+      },
+      "model": "Qwen/Qwen3-4B-Instruct-2507",
+      "model_revision": "cdbee75f17c01a7cc42f958dc650907174af0554",
+      "outcomes": 16384,
+      "rejected": 2853,
+      "source_evidence_sha256": "b61bb37cdb4adc56692137f6a3336fa0f6cdc475d0fe2b98f86d5e106d6961bb"
+    }
+  },
+  "rewrites": 32768,
+  "schema": "trellismark-rephrasing-public-export/v1",
+  "source_dataset": {
+    "commit": "c3ed277f6cb167bf8339e404828dcdc9099b04c4",
+    "repository": "xlr8harder/trellismark-qwen3-4b",
+    "revision": "v1.0.2"
+  },
+  "source_jobs": {
+    "path": "artifacts/trellismark-confirmation-rephrase-final-v1/jobs.jsonl",
+    "sha256": "054af84ffe4258f124c68b6c42ad715037b7806050c6b4cd3eb7406f46b5023e"
+  },
+  "sources": 16384
+}

results/rephrase-residual-carry-v1.csv

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:f4e5d94dbc8ed356fb2ba8509514b97db2f2557b417b6ced293ba1bd70f970cb
+size 11635385

rewrites/rewrites-00000-of-00005.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e6af0d57988c1d4af493724e30c02bdadb30620accd8cd75a526ee1d14cf4e1b
+size 7659197

rewrites/rewrites-00001-of-00005.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:2ead5fb369b85bc7bfd27760f6e114450cdeb8ba498cf911f48af0150bc5517a
+size 7711152

rewrites/rewrites-00002-of-00005.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:7ccda18aa5bf0651182e392e48e25782b682ab24edc31f4d99859f9148dfeb51
+size 7176675

rewrites/rewrites-00003-of-00005.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d1a1ca90d42de0f5ca13d150be417bf29a701a57cd9ad5b1c19f9d16a7c2ad7c
+size 7208343

rewrites/rewrites-00004-of-00005.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:9ad8a66613edaac4c5b0b7a74bc2113554ae8adf0573f38a1041ad5a11ae34b4
+size 807273

sources/sources-00000-of-00003.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:4bd6471f0f891fdec936188605d2be91c2d7b5d4cce3b941df8d88866448fd07
+size 5593052

sources/sources-00001-of-00003.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:308f3ee35a50ba330c119223765315d7684b5f4207126772dd67eefdff417e92
+size 5595457

sources/sources-00002-of-00003.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:067318ffc46ca0a1f949caa74574616c345566955ab237ad2e8324bd6babdf05
+size 317159