Add newly completed evaluation runs

README.md

@@ -1,6 +1,6 @@
 ---
 pretty_name: LLM Watermark Evaluation — Inputs and Generations (Qwen3.5-9B)
-license: cc-by-4.0
+license: [cc-by-4.0, cc-by-sa-4.0]
 language: [en, ar, zh, hi, ja]
 task_categories: [text-generation]
 tags: [watermarking, llm-watermarking, ai-generated-text, text-detection, robustness]
@@ -85,6 +85,26 @@ configs:
     path: data/none/benchmarks.parquet
   - split: sweet
     path: data/sweet/benchmarks.parquet
+  - split: kgw_delta1.0
+    path: data/kgw_delta1.0/benchmarks.parquet
+- config_name: documents
+  data_files:
+  - split: thicket
+    path: data/thicket/documents.parquet
+  - split: textseal
+    path: data/textseal/documents.parquet
+  - split: gumbelmax
+    path: data/gumbelmax/documents.parquet
+  - split: synthid
+    path: data/synthid/documents.parquet
+  - split: kgw
+    path: data/kgw/documents.parquet
+  - split: unigram
+    path: data/unigram/documents.parquet
+  - split: none
+    path: data/none/documents.parquet
+  - split: sweet
+    path: data/sweet/documents.parquet
 ---
 
 # LLM Watermark Evaluation — Inputs and Generations (Qwen3.5-9B)
@@ -107,17 +127,17 @@ one = load_dataset("RootAccess4Life/llm-watermark-eval-generations", "main", spl
 
 One split per watermark scheme in each of the configs below; the counts are rows.
 
-| scheme (split) | scheme and settings | main | dilution_answers | multilingual | benchmarks |
-|---|---|---|---|---|---|
-| `thicket` | Thicket (ours): keyed token-pair circle tilt with a likelihood price; κ = 6, price γ = 2 in answers (4 / 1 inside the reasoning delimiter), top-128 candidate shortlist, 256-label circle, first token and every 512-token restart neutral | 5,000 | 200 | 500 | 300 |
-| `textseal` | TextSeal (official code), mixing α = 0.1 | 5,000 | 200 | 500 | 300 |
-| `gumbelmax` | Gumbel-max / Aaronson (TextSeal-repo implementation) | 5,000 | 200 | 500 | 300 |
-| `synthid` | SynthID-Text, weighted tournament, depth 10 (TextSeal-repo implementation) | 5,000 | 200 | 500 | 300 |
-| `kgw` | KGW (Kirchenbauer et al. 2023): γ = 0.25, δ = 2.0, context width 1 | 5,000 | 200 | 500 | 300 |
-| `unigram` | Unigram (Zhao et al. 2024): fixed green list, γ = 0.25, δ = 2.0 | 5,000 | 200 | 500 | 300 |
-| `none` | No watermark (plain sampling; the unwatermarked reference) | 5,000 | 200 | 500 | 300 |
-| `sweet` | SWEET (Lee et al. 2024): entropy-gated KGW, γ = 0.25, δ = 2.0, entropy threshold 1.2 | 5,000 | 200 | 500 | 300 |
-| `kgw_delta1.0` | KGW, γ = 0.25, δ = 1.0 | 5,000 | 200 | 500 | – |
+| scheme (split) | scheme and settings | main | dilution_answers | multilingual | benchmarks | documents |
+|---|---|---|---|---|---|---|
+| `thicket` | Thicket (ours): keyed token-pair circle tilt with a likelihood price; κ = 6, price γ = 2 in answers (4 / 1 inside the reasoning delimiter), top-128 candidate shortlist, 256-label circle, first token and every 512-token restart neutral | 5,000 | 200 | 500 | 300 | 1,400 |
+| `textseal` | TextSeal (official code), mixing α = 0.1 | 5,000 | 200 | 500 | 300 | 1,400 |
+| `gumbelmax` | Gumbel-max / Aaronson (TextSeal-repo implementation) | 5,000 | 200 | 500 | 300 | 1,400 |
+| `synthid` | SynthID-Text, weighted tournament, depth 10 (TextSeal-repo implementation) | 5,000 | 200 | 500 | 300 | 1,400 |
+| `kgw` | KGW (Kirchenbauer et al. 2023): γ = 0.25, δ = 2.0, context width 1 | 5,000 | 200 | 500 | 300 | 1,400 |
+| `unigram` | Unigram (Zhao et al. 2024): fixed green list, γ = 0.25, δ = 2.0 | 5,000 | 200 | 500 | 300 | 1,400 |
+| `none` | No watermark (plain sampling; the unwatermarked reference) | 5,000 | 200 | 500 | 300 | 1,400 |
+| `sweet` | SWEET (Lee et al. 2024): entropy-gated KGW, γ = 0.25, δ = 2.0, entropy threshold 1.2 | 5,000 | 200 | 500 | 300 | 1,400 |
+| `kgw_delta1.0` | KGW, γ = 0.25, δ = 1.0 | 5,000 | 200 | 500 | 300 | – |
 
 Each scheme used one fixed key; the keys are released in [`keys.json`](keys.json) and explained in [`KEYS.md`](KEYS.md), so the detection results can be reproduced with the project's code.
 
@@ -126,25 +146,29 @@ Each scheme used one fixed key; the keys are released in [`keys.json`](keys.json
 | split | inputs | outputs | sampling |
 |---|---|---|---|
 | `main` | 1,000 English questions (ELI5, test split) | 5 independent answers per question (`sample` 0–4), at most 400 tokens | temperature 0.8, top-p 0.9; thinking disabled |
-| `dilution_answers` | the first 200 of those questions | 1 answer each; the evaluation embeds these answers into 2k–12k-token Wikipedia documents to test detection inside longer text | temperature 1.0, top-p 0.95 |
+| `dilution_answers` | the first 200 of those questions | 1 answer each; the evaluation embeds these answers into 2k–12k-token Wikipedia documents to test detection inside longer text (see `documents`) | temperature 1.0, top-p 0.95 |
 | `multilingual` | 100 questions in each of English, Arabic, Chinese, Hindi, Japanese | 1 answer each, with its reasoning trace (`reasoning`) and final answer (`answer`); the watermark covers both | temperature 0.7, top-p 0.95; at most 2,000 reasoning tokens, 4,096 tokens in total |
 | `benchmarks` | 300 questions from GSM8K, MATH-500, AIME, MMLU, ARC-Easy, ARC-Challenge, HellaSwag, WinoGrande, HumanEval and MBPP (30 each), with their reference answers | 1 answer each (reasoning trace and final answer), used to measure the effect of each watermark on task accuracy | temperature 0.6, top-p 0.95; at most 3,000 reasoning tokens, 4,096 tokens in total |
+| `documents` | the 200 `dilution_answers` texts and unwatermarked English Wikipedia host documents | 1,400 documents: each answer embedded whole in a 2k, 4k, 8k or 12k-token document (`dilute_*`) or cut into k = 2, 3 or 5 pieces scattered over an 8k-token document (`fragment_8k_k*`); `true_spans` marks the watermarked text | insertion points drawn with a fixed seed; lengths in generator tokens |
 
 ## Fields
 
 | field | meaning |
 |---|---|
 | `method` | watermark scheme (the config name) |
-| `model` | the generator that produced the text |
+| `model` | the generator that produced the text (in `documents`: the generator of the source answer) |
 | `sample` | sample index 0–4 (the five samples use different sampling seeds) |
-| `prompt_index` | position of the question in the evaluation's list |
+| `prompt_index` | position of the question in the evaluation's list (in `documents`: of the embedded answer, which `dilution_answers` holds under the same `prompt_id`) |
 | `prompt_id` | identifier of the question in its source dataset |
 | `question` | the question as written in the source dataset |
 | `prompt` | the exact string given to the model: the question inside the generator's chat template |
 | `language` | multilingual splits only |
 | `benchmark`, `kind`, `reference_answer` | `benchmarks` only: source benchmark, answer type and gold answer |
-| `text` | the generated text |
+| `text` | the generated text; in `documents` the whole document |
 | `reasoning`, `answer` | multilingual splits and `benchmarks`: `text` split at its first `</think>` (the answer stripped of surrounding whitespace); empty `reasoning` when the generator has no thinking mode |
+| `condition`, `length`, `k` | documents only: dilution / fragmentation condition, target document length in generator tokens, number of watermarked pieces |
+| `host_id` | documents only: the Wikipedia host document (consecutive English Wikipedia articles concatenated to the target length) |
+| `true_spans` | documents only: character offsets (`start`, `end`, Python string indices) of the watermarked pieces in `text` |
 | `n_tokens` | length of `text` in tokens of the generator's tokenizer |
 
 ## Generation
@@ -169,6 +193,8 @@ Generated texts: CC BY 4.0. They are outputs of the generator named in `model`,
 model's terms:
 - `Qwen/Qwen3.5-9B`: subject to the Qwen3.5 model license (see the model card).
 
+**`documents` / `c4_documents`: CC BY-SA 4.0.** These splits contain text from English Wikipedia (`wikimedia/wikipedia`, dump 20231101.en; © Wikipedia contributors, https://en.wikipedia.org), redistributed under CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/); the generated passages they embed are marked by `true_spans`. All other splits: CC BY 4.0.
+
 Inputs keep the licenses of their sources: ELI5 (Reddit questions, as distributed in the `Pavithree/eli5` Hugging
 Face dataset; also the English multilingual questions), the multilingual question pool (`shanearora/CaLMQA`, CC BY 4.0),
 GSM8K (MIT), MATH-500 (MIT), AIME (public competition problems), MMLU (MIT), ARC (CC BY-SA 4.0), HellaSwag (MIT),

data/gumbelmax/documents.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:edad60606a5c34ca89fea2707dc3914fd18969d9b34cbd13265c735b018871c3
+size 16834430

data/kgw/documents.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:0def5edb613b879b0f7e492a889b0f563313e53c93bd86433d551506471d6ddf
+size 16850516

data/kgw_delta1.0/benchmarks.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:e2d608c0f2f17a15ccddd0bf2010ff552feaee3a8ecc1ac581575033bb59daa0
+size 1368372

data/none/documents.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:004a122d61f0bffa57dd0e2cf222f3650dbb49a536d1b567cc08587b06f9f6be
+size 16843246

data/sweet/documents.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d878dc6037031f66c89d1b83149f2cf4f48aa474f5037040ca5e43f87050a16f
+size 16842129

data/synthid/documents.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d8fe9bd348953aa4852f7a6b9c9e6167decc07dd0620446984dbb8a2f70f0bc1
+size 16842142

data/textseal/documents.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:5a43e0e6876407744934e67aac90b779fd464c88569c869c7b996bc16260d977
+size 16846606

data/thicket/documents.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:1f7719077350e374db4bfcbeab093c476555a43f855c10679f50edace96fc2ee
+size 16836284

data/unigram/documents.parquet

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:b0f0454fa39c98236bf21e9aa7eb664f4368282fd4cf553f1069a7fc03c13dbd
+size 16833857