G-reen/fastdetector-test-stat-test

Auto-generated FastDetector dataset

Fastdetector Test Stat

Best detector

Giga EditLens Llama-3.2-3B Score

0.6359 TPR @ 1% FPR

Hardest prompt subset

rewrite

0.2464 max detector TPR @ 1% FPR

Hardest generator config

claude-opus-5 (Temp: Unknown)

0.4089 max detector TPR @ 1% FPR

15,695

rows

10

generator configs

4

prompt subsets

15

detectors

01Leaderboard02Model analytics03Distances04Appendix

01

Detector leaderboard

Score-based detectors ranked by overall AUROC. Thresholds are placed exactly on every human score, so TPR is reported at a true 1% and 0.1% false positive rate. Open a row for its threshold sweep, its TPR as AI rows with low cosdist are dropped, and its score distributions.

01Giga EditLens Llama-3.2-3B ScoreN31,390AUROC0.9040TPR @ 1% FPR0.6359TPR @ 0.1% FPR0.4950⌄

1% FPR threshold 0.04330.1% FPR threshold 0.1831

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

02EditLens Llama-3.2-3B ScoreN31,390AUROC0.8805TPR @ 1% FPR0.5710TPR @ 0.1% FPR0.3976⌄

1% FPR threshold 0.20160.1% FPR threshold 0.3730

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

03Giga EditLens Roberta ScoreN31,390AUROC0.8677TPR @ 1% FPR0.5543TPR @ 0.1% FPR0.4872⌄

1% FPR threshold 0.12390.1% FPR threshold 0.2631

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

04EditLens Roberta-Large ScoreN31,390AUROC0.8550TPR @ 1% FPR0.4431TPR @ 0.1% FPR0.1832⌄

1% FPR threshold 0.30800.1% FPR threshold 0.7948

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

05BinocularsN31,390AUROC0.6064TPR @ 1% FPR0.0670TPR @ 0.1% FPR0.0182⌄

1% FPR threshold 0.95030.1% FPR threshold 0.9907

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

06Top-p Outliers (Llama-3.2-3B)N31,390AUROC0.5697TPR @ 1% FPR0.0503TPR @ 0.1% FPR0.0022⌄

1% FPR threshold 0.02190.1% FPR threshold 0.0071

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

07Entropy (Llama-3.2-3B-Instruct)N31,390AUROC0.5345TPR @ 1% FPR0.0187TPR @ 0.1% FPR0.0041⌄

1% FPR threshold 1.46500.1% FPR threshold 1.1720

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

08Perplexity (Llama-3.2-3B-Instruct)N31,390AUROC0.5234TPR @ 1% FPR0.0190TPR @ 0.1% FPR0.0038⌄

1% FPR threshold 5.25180.1% FPR threshold 3.6443

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

09Top-k Outliers (Llama-3.2-3B-Instruct)N31,390AUROC0.5109TPR @ 1% FPR0.0247TPR @ 0.1% FPR0.0036⌄

1% FPR threshold 0.03120.1% FPR threshold 0.0161

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

10Top-p Outliers (Llama-3.2-3B-Instruct)N31,390AUROC0.5025TPR @ 1% FPR0.0273TPR @ 0.1% FPR0.0038⌄

1% FPR threshold 0.03280.1% FPR threshold 0.0234

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

11Perplexity (Llama-3.2-3B)N31,390AUROC0.4873TPR @ 1% FPR0.0091TPR @ 0.1% FPR0.0004⌄

1% FPR threshold 3.80030.1% FPR threshold 1.3646

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

12FastDetectGPT (Llama-3.2-3B-Instruct)N31,390AUROC0.4827TPR @ 1% FPR0.0146TPR @ 0.1% FPR0.0014⌄

1% FPR threshold 2.20840.1% FPR threshold 5.4559

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

13Top-k Outliers (Llama-3.2-3B)N31,390AUROC0.4806TPR @ 1% FPR0.0129TPR @ 0.1% FPR0.0006⌄

1% FPR threshold 0.01960.1% FPR threshold 0.0043

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

14Entropy (Llama-3.2-3B)N31,390AUROC0.4751TPR @ 1% FPR0.0072TPR @ 0.1% FPR0.0005⌄

1% FPR threshold 1.35260.1% FPR threshold 0.3718

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

15FastDetectGPT (Llama-3.2-3B)N31,390AUROC0.4595TPR @ 1% FPR0.0250TPR @ 0.1% FPR0.0015⌄

1% FPR threshold -2.68350.1% FPR threshold -4.6137

TPR and FPR at every threshold · both FPR operating points markedOpen full size ↗

AI rows with cosdist below the cutoff are dropped · human rows and thresholds are unchangedOpen full size ↗

Score distributions

Overall · human onlyOpen full size ↗

Overall · AI onlyOpen full size ↗

AI texts · one envelope per prompt subsetOpen full size ↗

AI texts · one envelope per generator configOpen full size ↗

02

Model-Specific Analytics

Select a model, then a split; the split's charts appear below it. Topic × format compares human scores, AI scores and their unsigned gap; generator × prompt category shows only the AI side, since neither changes the human text; specific prompt ranks every first message by difficulty.

Select a model

01Giga EditLens Llama-3.2-3B ScoreAUROC 0.9040 · TPR@1% 0.6359⌄

Select a split · Giga EditLens Llama-3.2-3B Score

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched. The final text must have the format <<format>> and topic <<topic>>. rewrite · single-turn	9	0.5802	0.0000	0.0000	0.0044	0.0086	0.0043	0.0782	0.0049
Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched. The final text must have the topic <<topic>>. rewrite · single-turn	18	0.6173	0.0000	0.0000	0.0048	0.0057	0.0015	0.0675	-0.0000
Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound. The final text must have the format <<format>> and topic <<topic>>. rewrite · single-turn	18	0.6019	0.0000	0.0000	0.0032	0.0057	0.0028	0.0954	0.0022
Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound. The final text must have the format <<format>>. rewrite · single-turn	20	0.6125	0.0000	0.0000	0.0032	0.0042	0.0014	0.1036	0.0018
Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound. The final text must have the topic <<topic>>. rewrite · single-turn	9	0.5926	0.0000	0.0000	0.0033	0.0041	0.0012	0.0914	0.0020
Edit this text by removing exactly one to three sentences from it (the least necessary), you may only make minimal modifications to adjacent sentences for better flow. The final text must have the format <<format>>. rewrite · single-turn	15	0.7156	0.0000	0.0000	0.0054	0.0084	0.0032	0.0902	0.0073
Edit this text by removing exactly one to three sentences from it (the least necessary), you may only make minimal modifications to adjacent sentences for better flow. The final text must have the topic <<topic>>. rewrite · single-turn	11	0.5207	0.0000	0.0000	0.0035	0.0038	0.0017	0.1127	0.0150
Edit this text by removing five to ten percent of the words from it (the least necessary), without changing anything else. The final text must have the format <<format>>. rewrite · single-turn	14	0.6888	0.0000	0.0000	0.0050	0.0085	0.0038	0.1057	-0.0002
Edit this text by removing five to ten percent of the words from it (the least necessary), without changing anything else. The final text must have the topic <<topic>>. rewrite · single-turn	13	0.6864	0.0000	0.0000	0.0023	0.0033	0.0010	0.0662	-0.0002
Edit this text by removing one section from it (the least necessary), you may only make minimal modifications to adjacent sections for better flow. The final text must have the topic <<topic>>. rewrite · single-turn	22	0.5310	0.0000	0.0000	0.0035	0.0037	0.0013	0.1421	0.0150
Edit this text by replacing all existing transition words and phrases (e.g., 'however', 'therefore', 'in addition') with different, synonymous transitions, making no other changes. The final text must have the format <<format>> and topic <<topic>>. rewrite · single-turn	17	0.5986	0.0000	0.0000	0.0032	0.0051	0.0025	0.0833	0.0003
Edit this text by replacing all existing transition words and phrases (e.g., 'however', 'therefore', 'in addition') with different, synonymous transitions, making no other changes. The final text must have the format <<format>>. rewrite · single-turn	10	0.6300	0.0000	0.0000	0.0040	0.0070	0.0031	0.0772	0.0006

02EditLens Llama-3.2-3B ScoreAUROC 0.8805 · TPR@1% 0.5710⌄

Select a split · EditLens Llama-3.2-3B Score

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched. The final text must have the format <<format>> and topic <<topic>>. rewrite · single-turn	9	0.5432	0.0000	0.0000	0.0287	0.0360	0.0180	0.0782	0.0049
Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched. The final text must have the format <<format>>. rewrite · single-turn	12	0.6319	0.0000	0.0000	0.0311	0.0392	0.0322	0.1024	0.0140
Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched. The final text must have the topic <<topic>>. rewrite · single-turn	18	0.5926	0.0000	0.0000	0.0259	0.0381	0.0148	0.0675	-0.0000
Edit this text by inserting exactly one to three sentences (in appropriate locations), you may only make minimal modifications to transitions in adjacent sentences for better flow. The final text must have the format <<format>>. rewrite · single-turn	21	0.6508	0.0000	0.0000	0.0174	0.0491	0.0318	0.0941	0.0149
Edit this text by inserting twenty to thirty words (in appropriate locations), without changing anything else. The final text must have the format <<format>> and topic <<topic>>. rewrite · single-turn	16	0.6836	0.0000	0.0000	0.0194	0.0381	0.0200	0.0725	0.0084
Edit this text by inserting twenty to thirty words (in appropriate locations), without changing anything else. The final text must have the topic <<topic>>. rewrite · single-turn	17	0.6886	0.0000	0.0000	0.0319	0.0498	0.0180	0.0686	0.0214
Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound. The final text must have the format <<format>> and topic <<topic>>. rewrite · single-turn	18	0.5957	0.0000	0.0000	0.0315	0.0443	0.0190	0.0954	0.0022
Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound. The final text must have the format <<format>>. rewrite · single-turn	20	0.5700	0.0000	0.0000	0.0226	0.0300	0.0115	0.1036	0.0018
Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound. The final text must have the topic <<topic>>. rewrite · single-turn	9	0.5679	0.0000	0.0000	0.0389	0.0457	0.0077	0.0914	0.0020
Edit this text by removing exactly one to three sentences from it (the least necessary), you may only make minimal modifications to adjacent sentences for better flow. The final text must have the format <<format>> and topic <<topic>>. rewrite · single-turn	15	0.5911	0.0000	0.0000	0.0150	0.0315	0.0170	0.0973	0.0182
Edit this text by removing exactly one to three sentences from it (the least necessary), you may only make minimal modifications to adjacent sentences for better flow. The final text must have the format <<format>>. rewrite · single-turn	15	0.6222	0.0000	0.0000	0.0160	0.0222	0.0085	0.0902	0.0073
Edit this text by removing exactly one to three sentences from it (the least necessary), you may only make minimal modifications to adjacent sentences for better flow. The final text must have the topic <<topic>>. rewrite · single-turn	11	0.5620	0.0000	0.0000	0.0263	0.0256	0.0058	0.1127	0.0150

03Giga EditLens Roberta ScoreAUROC 0.8677 · TPR@1% 0.5543⌄

Select a split · Giga EditLens Roberta Score

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched. The final text must have the format <<format>>. rewrite · single-turn	12	0.5347	0.0000	0.0000	0.0064	0.0066	0.0024	0.1024	0.0140
Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound. The final text must have the topic <<topic>>. rewrite · single-turn	9	0.5432	0.0000	0.0000	0.0040	0.0042	0.0005	0.0914	0.0020
Edit this text by removing exactly one to three sentences from it (the least necessary), you may only make minimal modifications to adjacent sentences for better flow. The final text must have the topic <<topic>>. rewrite · single-turn	11	0.5702	0.0000	0.0000	0.0045	0.0042	0.0018	0.1127	0.0150
Edit this text by removing five to ten percent of the words from it (the least necessary), without changing anything else. The final text must have the format <<format>> and topic <<topic>>. rewrite · single-turn	20	0.6450	0.0000	0.0000	0.0057	0.0173	0.0118	0.1138	0.0072
Edit this text by removing five to ten percent of the words from it (the least necessary), without changing anything else. The final text must have the format <<format>>. rewrite · single-turn	14	0.6633	0.0000	0.0000	0.0050	0.0077	0.0027	0.1057	-0.0002
Edit this text by removing five to ten percent of the words from it (the least necessary), without changing anything else. The final text must have the topic <<topic>>. rewrite · single-turn	13	0.6213	0.0000	0.0000	0.0042	0.0055	0.0015	0.0662	-0.0002
Edit this text by removing one section from it (the least necessary), you may only make minimal modifications to adjacent sections for better flow. The final text must have the topic <<topic>>. rewrite · single-turn	22	0.5444	0.0000	0.0000	0.0046	0.0057	0.0017	0.1421	0.0150
Edit this text by replacing all existing transition words and phrases (e.g., 'however', 'therefore', 'in addition') with different, synonymous transitions, making no other changes. The final text must have the format <<format>> and topic <<topic>>. rewrite · single-turn	17	0.5363	0.0000	0.0000	0.0035	0.0046	0.0013	0.0833	0.0003
Edit this text by replacing all existing transition words and phrases (e.g., 'however', 'therefore', 'in addition') with different, synonymous transitions, making no other changes. The final text must have the format <<format>>. rewrite · single-turn	10	0.5200	0.0000	0.0000	0.0130	0.0112	0.0033	0.0772	0.0006
Edit this text by replacing all existing transition words and phrases (e.g., 'however', 'therefore', 'in addition') with different, synonymous transitions, making no other changes. The final text must have the topic <<topic>>. rewrite · single-turn	8	0.5000	0.0000	0.0000	0.0071	0.0068	0.0025	0.1952	0.0487
Edit this text by replacing exactly one to three sentences, without changing anything else. The final text must have the format <<format>> and topic <<topic>>. rewrite · single-turn	16	0.5449	0.0000	0.0000	0.0059	0.0064	0.0005	0.0871	0.0105
Edit this text by replacing exactly one to three sentences, without changing anything else. The final text must have the topic <<topic>>. rewrite · single-turn	12	0.5174	0.0000	0.0000	0.0067	0.0093	0.0033	0.0880	0.0171

04EditLens Roberta-Large ScoreAUROC 0.8550 · TPR@1% 0.4431⌄

Select a split · EditLens Roberta-Large Score

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched. The final text must have the format <<format>> and topic <<topic>>. rewrite · single-turn	9	0.4938	0.0000	0.0000	0.0595	0.0596	0.0030	0.0782	0.0049
Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched. The final text must have the format <<format>>. rewrite · single-turn	12	0.5556	0.0000	0.0000	0.0383	0.0414	0.0136	0.1024	0.0140
Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched. The final text must have the topic <<topic>>. rewrite · single-turn	18	0.6327	0.0000	0.0000	0.0338	0.0499	0.0172	0.0675	-0.0000
Edit this text by completely rewriting the topic sentence of every paragraph, while preserving the supporting sentences in those paragraphs verbatim. The final text must have the topic <<topic>>. rewrite · single-turn	15	0.7156	0.0000	0.0000	0.0268	0.0660	0.0415	0.3260	0.0983
Edit this text by inserting a single, highly engaging introductory sentence at the very beginning, without modifying any of the original text that follows. The final text must have the topic <<topic>>. rewrite · single-turn	24	0.6267	0.0000	0.0000	0.0481	0.0817	0.0337	0.0508	0.0327
Edit this text by inserting a singular new section (in an appropriate locations), you may only make minimal modifications to transitions in adjacent sections for better flow. The final text must have the format <<format>> and topic <<topic>>. rewrite · single-turn	20	0.7450	0.0000	0.0000	0.0469	0.1103	0.0636	0.1784	0.0338
Edit this text by inserting a singular new section (in an appropriate locations), you may only make minimal modifications to transitions in adjacent sections for better flow. The final text must have the format <<format>>. rewrite · single-turn	12	0.7049	0.0000	0.0000	0.0626	0.1045	0.0434	0.1629	0.0161
Edit this text by inserting exactly one to three sentences (in appropriate locations), you may only make minimal modifications to transitions in adjacent sentences for better flow. The final text must have the topic <<topic>>. rewrite · single-turn	15	0.6600	0.0000	0.0000	0.0237	0.0467	0.0232	0.0694	0.0214
Edit this text by inserting twenty to thirty words (in appropriate locations), without changing anything else. The final text must have the topic <<topic>>. rewrite · single-turn	17	0.6349	0.0000	0.0000	0.0351	0.0505	0.0156	0.0686	0.0214
Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound. The final text must have the format <<format>>. rewrite · single-turn	20	0.5400	0.0000	0.0000	0.0410	0.0381	0.0157	0.1036	0.0018
Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound. The final text must have the topic <<topic>>. rewrite · single-turn	9	0.5185	0.0000	0.0000	0.0513	0.0501	0.0107	0.0914	0.0020
Edit this text by removing exactly one to three sentences from it (the least necessary), you may only make minimal modifications to adjacent sentences for better flow. The final text must have the format <<format>> and topic <<topic>>. rewrite · single-turn	15	0.6444	0.0000	0.0000	0.0250	0.0345	0.0222	0.0973	0.0182

05BinocularsAUROC 0.6064 · TPR@1% 0.0670⌄

Select a split · Binoculars

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Adapt this text for a 5th-grade reading level. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	8	0.7031	0.0000	0.0000	0.8406	0.8622	0.0294	0.6715	0.1082
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	12	0.6875	0.0000	0.0000	0.8164	0.8955	0.0886	0.7114	0.1210
Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>>. revise · single-turn	9	0.6790	0.0000	0.0000	0.8427	0.8699	0.0406	0.7067	0.1031
Change the tone of this text to be more academic and scholarly. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	14	0.5153	0.0000	0.0000	0.8460	0.8515	0.0275	0.7816	0.1009
Change the tone of this text to be more academic and scholarly. The final text must have the format <<format>>. revise · single-turn	12	0.4444	0.0000	0.0000	0.8608	0.8589	0.0259	0.7366	0.0965
Change the tone of this text to be more academic and scholarly. The final text must have the topic <<topic>>. revise · single-turn	8	0.5938	0.0000	0.0000	0.8493	0.8610	0.0294	0.7892	0.2192
Change the tone of this text to be more analytical and logical. The final text must have the topic <<topic>>. revise · single-turn	11	0.6116	0.0000	0.0000	0.7953	0.8471	0.0629	0.7514	0.1159
Change the tone of this text to be more conversational and approachable. The final text must have the topic <<topic>>. revise · single-turn	11	0.5702	0.0000	0.0000	0.8626	0.8650	0.0171	0.6663	0.0460
Change the tone of this text to be more empathetic and understanding. The final text must have the format <<format>>. revise · single-turn	14	0.6173	0.0000	0.0000	0.8238	0.8721	0.0690	0.5995	0.0808
Change the tone of this text to be more empathetic and understanding. The final text must have the topic <<topic>>. revise · single-turn	6	0.5556	0.0000	0.0000	0.8152	0.8637	0.0598	0.6320	0.0727
Change the tone of this text to be more enthusiastic and energetic. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	15	0.4756	0.0000	0.0000	0.8603	0.8524	0.0478	0.6233	0.0637
Change the tone of this text to be more enthusiastic and energetic. The final text must have the topic <<topic>>. revise · single-turn	4	0.6875	0.0000	0.0000	0.8418	0.8508	0.0391	0.5004	0.1208

06Top-p Outliers (Llama-3.2-3B)AUROC 0.5697 · TPR@1% 0.0503⌄

Select a split · Top-p Outliers (Llama-3.2-3B)

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>>. revise · single-turn	13	0.6982	0.0000	0.0000	0.0428	0.0363	0.0103	0.7209	0.1117
Adapt this text for a teenage audience. revise · single & two-turn	63	0.5629	0.0000	0.0000	0.0415	0.0400	0.0103	0.7319	0.1237
Adapt this text for a teenage audience. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	11	0.6446	0.0000	0.0000	0.0387	0.0354	0.0092	0.6970	0.0923
Adapt this text for a teenage audience. The final text must have the format <<format>>. revise · single-turn	9	0.9383	0.0000	0.0000	0.0506	0.0351	0.0156	0.6764	0.0705
Adapt this text for a teenage audience. The final text must have the topic <<topic>>. revise · single-turn	14	0.6684	0.0000	0.0000	0.0411	0.0369	0.0085	0.6562	0.1132
Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>>. revise · single-turn	9	0.8519	0.0000	0.0000	0.0457	0.0370	0.0104	0.7067	0.1031
Adapt this text for social media sharing. revise · single & two-turn	52	0.5643	0.0000	0.0000	0.0421	0.0408	0.0113	0.7845	0.1244
Adapt this text for social media sharing. The final text must have the format <<format>>. revise · single-turn	11	0.5620	0.0000	0.0000	0.0460	0.0468	0.0073	0.7383	0.1500
Adapt this text for social media sharing. The final text must have the topic <<topic>>. revise · single-turn	13	0.4320	0.0000	0.0000	0.0392	0.0406	0.0108	0.7792	0.1240
Change the tone of this text to be more academic and scholarly. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	14	0.6276	0.0000	0.0000	0.0435	0.0379	0.0092	0.7816	0.1009
Change the tone of this text to be more academic and scholarly. The final text must have the format <<format>>. revise · single-turn	12	0.7361	0.0000	0.0000	0.0410	0.0336	0.0087	0.7366	0.0965
Change the tone of this text to be more academic and scholarly. The final text must have the topic <<topic>>. revise · single-turn	8	0.6875	0.0000	0.0000	0.0412	0.0337	0.0101	0.7892	0.2192

07Entropy (Llama-3.2-3B-Instruct)AUROC 0.5345 · TPR@1% 0.0187⌄

Select a split · Entropy (Llama-3.2-3B-Instruct)

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Adapt this text for a 5th-grade reading level. revise · single & two-turn	48	0.6944	0.0000	0.0000	2.5127	2.2511	0.4071	0.7738	0.1348
Adapt this text for a 5th-grade reading level. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	8	0.7812	0.0000	0.0000	2.5667	2.1258	0.4500	0.6715	0.1082
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	12	0.7153	0.0000	0.0000	2.4028	2.0025	0.4632	0.7114	0.1210
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>>. revise · single-turn	13	0.7692	0.0000	0.0000	2.5786	2.1337	0.4484	0.7209	0.1117
Adapt this text for a teenage audience. revise · single & two-turn	63	0.5770	0.0000	0.0000	2.5388	2.4271	0.3452	0.7319	0.1237
Adapt this text for a teenage audience. The final text must have the topic <<topic>>. revise · single-turn	14	0.7041	0.0000	0.0000	2.5357	2.1864	0.4199	0.6562	0.1132
Adapt this text for an elderly adult with limited knowledge of modern day affairs. revise · single & two-turn	62	0.5981	0.0000	0.0000	2.5038	2.3436	0.3378	0.7264	0.1338
Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>>. revise · single-turn	9	0.7407	0.0000	0.0000	2.4414	2.1424	0.3797	0.7067	0.1031
Adapt this text for social media sharing. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	14	0.6020	0.0000	0.0000	2.4232	2.2368	0.2899	0.7022	0.1296
Adapt this text for social media sharing. The final text must have the format <<format>>. revise · single-turn	11	0.7190	0.0000	0.0000	2.4975	2.2557	0.3360	0.7383	0.1500
Adapt this text for social media sharing. The final text must have the topic <<topic>>. revise · single-turn	13	0.6154	0.0000	0.0000	2.4276	2.2817	0.4448	0.7792	0.1240
Change the tone of this text to be more academic and scholarly. The final text must have the format <<format>>. revise · single-turn	12	0.4167	0.0000	0.0000	2.4815	2.6055	0.2759	0.7366	0.0965

08Perplexity (Llama-3.2-3B-Instruct)AUROC 0.5234 · TPR@1% 0.0190⌄

Select a split · Perplexity (Llama-3.2-3B-Instruct)

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Adapt this text for a lay audience with no technical background. revise · single & two-turn	54	0.6828	0.0000	0.0000	17.6196	12.3657	6.6714	0.7643	0.1436
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	12	0.6806	0.0000	0.0000	16.0092	9.5345	6.5983	0.7114	0.1210
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>>. revise · single-turn	13	0.7219	0.0000	0.0000	18.0796	11.2286	7.1788	0.7209	0.1117
Adapt this text for a teenage audience. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	11	0.6033	0.0000	0.0000	16.1392	13.8798	3.5450	0.6970	0.0923
Adapt this text for an elderly adult with limited knowledge of modern day affairs. revise · single & two-turn	62	0.5619	0.0000	0.0000	15.5077	13.7873	4.6786	0.7264	0.1338
Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>>. revise · single-turn	9	0.7778	0.0000	0.0000	14.4519	10.1888	5.1213	0.7067	0.1031
Adapt this text for social media sharing. revise · single & two-turn	52	0.4630	0.0000	0.0000	15.8323	16.0909	4.7022	0.7845	0.1244
Adapt this text for social media sharing. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	14	0.5969	0.0000	0.0000	16.6637	12.9454	5.7038	0.7022	0.1296
Adapt this text for social media sharing. The final text must have the format <<format>>. revise · single-turn	11	0.5455	0.0000	0.0000	14.4287	14.0512	4.7856	0.7383	0.1500
Adapt this text for social media sharing. The final text must have the topic <<topic>>. revise · single-turn	13	0.3373	0.0000	0.0000	13.4431	15.1373	4.5788	0.7792	0.1240
Change the tone of this text to be more academic and scholarly. The final text must have the format <<format>>. revise · single-turn	12	0.4792	0.0000	0.0000	14.8814	15.8015	3.9188	0.7366	0.0965
Change the tone of this text to be more academic and scholarly. The final text must have the topic <<topic>>. revise · single-turn	8	0.4531	0.0000	0.0000	15.8208	15.2101	5.9025	0.7892	0.2192

09Top-k Outliers (Llama-3.2-3B-Instruct)AUROC 0.5109 · TPR@1% 0.0247⌄

Select a split · Top-k Outliers (Llama-3.2-3B-Instruct)

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Adapt this text for a 5th-grade reading level. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	8	0.7969	0.0000	0.0000	0.0939	0.0640	0.0318	0.6715	0.1082
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>>. revise · single-turn	13	0.7574	0.0000	0.0000	0.1033	0.0686	0.0363	0.7209	0.1117
Adapt this text for a teenage audience. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	11	0.6116	0.0000	0.0000	0.1008	0.0921	0.0264	0.6970	0.0923
Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>>. revise · single-turn	9	0.8148	0.0000	0.0000	0.1005	0.0664	0.0404	0.7067	0.1031
Adapt this text for social media sharing. revise · single & two-turn	52	0.4118	0.0000	0.0000	0.0959	0.1033	0.0245	0.7845	0.1244
Adapt this text for social media sharing. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	14	0.5255	0.0000	0.0000	0.0955	0.0878	0.0258	0.7022	0.1296
Adapt this text for social media sharing. The final text must have the format <<format>>. revise · single-turn	11	0.4050	0.0000	0.0000	0.0855	0.0931	0.0188	0.7383	0.1500
Adapt this text for social media sharing. The final text must have the topic <<topic>>. revise · single-turn	13	0.2781	0.0000	0.0000	0.0903	0.1086	0.0289	0.7792	0.1240
Change the tone of this text to be more academic and scholarly. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	14	0.3827	0.0000	0.0000	0.0924	0.0994	0.0238	0.7816	0.1009
Change the tone of this text to be more academic and scholarly. The final text must have the format <<format>>. revise · single-turn	12	0.4306	0.0000	0.0000	0.0942	0.1045	0.0237	0.7366	0.0965
Change the tone of this text to be more academic and scholarly. The final text must have the topic <<topic>>. revise · single-turn	8	0.3906	0.0000	0.0000	0.0929	0.1053	0.0307	0.7892	0.2192
Change the tone of this text to be more analytical and logical. revise · single & two-turn	55	0.4555	0.0000	0.0000	0.0981	0.1030	0.0271	0.7751	0.1561

10Top-p Outliers (Llama-3.2-3B-Instruct)AUROC 0.5025 · TPR@1% 0.0273⌄

Select a split · Top-p Outliers (Llama-3.2-3B-Instruct)

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Adapt this text for a 5th-grade reading level. revise · single & two-turn	48	0.4740	0.0000	0.0000	0.0558	0.0566	0.0095	0.7738	0.1348
Adapt this text for a 5th-grade reading level. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	8	0.4688	0.0000	0.0000	0.0525	0.0534	0.0083	0.6715	0.1082
Adapt this text for a 5th-grade reading level. The final text must have the topic <<topic>>. revise · single-turn	13	0.4024	0.0000	0.0000	0.0572	0.0626	0.0194	0.7871	0.1714
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	12	0.4931	0.0000	0.0000	0.0591	0.0610	0.0100	0.7114	0.1210
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>>. revise · single-turn	13	0.5089	0.0000	0.0000	0.0558	0.0565	0.0112	0.7209	0.1117
Adapt this text for a lay audience with no technical background. The final text must have the topic <<topic>>. revise · single-turn	16	0.4570	0.0000	0.0000	0.0593	0.0611	0.0132	0.6803	0.1203
Adapt this text for a teenage audience. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	11	0.5620	0.0000	0.0000	0.0540	0.0518	0.0061	0.6970	0.0923
Adapt this text for a teenage audience. The final text must have the topic <<topic>>. revise · single-turn	14	0.2857	0.0000	0.0000	0.0508	0.0631	0.0125	0.6562	0.1132
Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>>. revise · single-turn	9	0.7778	0.0000	0.0000	0.0569	0.0492	0.0094	0.7067	0.1031
Adapt this text for social media sharing. revise · single & two-turn	52	0.3467	0.0000	0.0000	0.0544	0.0630	0.0150	0.7845	0.1244
Adapt this text for social media sharing. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	14	0.3673	0.0000	0.0000	0.0573	0.0632	0.0134	0.7022	0.1296
Adapt this text for social media sharing. The final text must have the format <<format>>. revise · single-turn	11	0.2562	0.0000	0.0000	0.0538	0.0748	0.0228	0.7383	0.1500

11Perplexity (Llama-3.2-3B)AUROC 0.4873 · TPR@1% 0.0091⌄

Select a split · Perplexity (Llama-3.2-3B)

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Adapt this text for a 5th-grade reading level. revise · single & two-turn	48	0.6141	0.0000	0.0000	11.8004	10.3328	3.7892	0.7738	0.1348
Adapt this text for a 5th-grade reading level. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	8	0.7656	0.0000	0.0000	12.4898	8.4790	4.0908	0.6715	0.1082
Adapt this text for a 5th-grade reading level. The final text must have the topic <<topic>>. revise · single-turn	13	0.7988	0.0000	0.0000	15.4699	10.9513	5.7348	0.7871	0.1714
Adapt this text for a lay audience with no technical background. revise · single & two-turn	54	0.6313	0.0000	0.0000	13.1474	10.2631	4.5008	0.7643	0.1436
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	12	0.6528	0.0000	0.0000	11.8051	7.9813	4.4912	0.7114	0.1210
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>>. revise · single-turn	13	0.7041	0.0000	0.0000	13.2378	9.1447	4.5290	0.7209	0.1117
Adapt this text for a teenage audience. revise · single & two-turn	63	0.5077	0.0000	0.0000	12.0871	12.2266	3.5824	0.7319	0.1237
Adapt this text for a teenage audience. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	11	0.5207	0.0000	0.0000	11.9794	11.2162	2.6308	0.6970	0.0923
Adapt this text for a teenage audience. The final text must have the format <<format>>. revise · single-turn	9	0.7407	0.0000	0.0000	13.6140	9.2517	4.4697	0.6764	0.0705
Adapt this text for a teenage audience. The final text must have the topic <<topic>>. revise · single-turn	14	0.5918	0.0000	0.0000	12.1179	10.2096	3.2666	0.6562	0.1132
Adapt this text for an elderly adult with limited knowledge of modern day affairs. revise · single & two-turn	62	0.5260	0.0000	0.0000	11.8059	11.2468	3.4307	0.7264	0.1338
Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>>. revise · single-turn	9	0.7160	0.0000	0.0000	10.9873	8.5971	3.2846	0.7067	0.1031

12FastDetectGPT (Llama-3.2-3B-Instruct)AUROC 0.4827 · TPR@1% 0.0146⌄

Select a split · FastDetectGPT (Llama-3.2-3B-Instruct)

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Adapt this text for a 5th-grade reading level. revise · single & two-turn	48	0.3971	0.0000	0.0000	-1.7428	-2.3858	1.4394	0.7738	0.1348
Adapt this text for a 5th-grade reading level. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	8	0.3594	0.0000	0.0000	-1.0936	-1.8732	1.3594	0.6715	0.1082
Adapt this text for a 5th-grade reading level. The final text must have the topic <<topic>>. revise · single-turn	13	0.1953	0.0000	0.0000	-1.7912	-3.2474	1.6244	0.7871	0.1714
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	12	0.4653	0.0000	0.0000	-2.5364	-2.7945	1.1319	0.7114	0.1210
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>>. revise · single-turn	13	0.3846	0.0000	0.0000	-2.0482	-2.7817	1.3008	0.7209	0.1117
Adapt this text for a lay audience with no technical background. The final text must have the topic <<topic>>. revise · single-turn	16	0.3828	0.0000	0.0000	-2.3219	-3.1766	1.6299	0.6803	0.1203
Adapt this text for a teenage audience. revise · single & two-turn	63	0.4482	0.0000	0.0000	-1.6502	-1.9644	1.7214	0.7319	0.1237
Adapt this text for a teenage audience. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	11	0.5620	0.0000	0.0000	-2.2489	-2.2397	1.1577	0.6970	0.0923
Adapt this text for a teenage audience. The final text must have the format <<format>>. revise · single-turn	9	0.4444	0.0000	0.0000	-1.1346	-2.0499	1.2788	0.6764	0.0705
Adapt this text for a teenage audience. The final text must have the topic <<topic>>. revise · single-turn	14	0.2500	0.0000	0.0000	-1.0898	-3.0366	2.0619	0.6562	0.1132
Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	13	0.3195	0.0000	0.0000	-1.4448	-2.1430	1.0665	0.7414	0.1845
Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>>. revise · single-turn	9	0.5802	0.0000	0.0000	-2.2551	-1.8167	1.3863	0.7067	0.1031

13Top-k Outliers (Llama-3.2-3B)AUROC 0.4806 · TPR@1% 0.0129⌄

Select a split · Top-k Outliers (Llama-3.2-3B)

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Adapt this text for a 5th-grade reading level. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	8	0.8281	0.0000	0.0000	0.0770	0.0495	0.0275	0.6715	0.1082
Adapt this text for a lay audience with no technical background. revise · single & two-turn	54	0.6252	0.0000	0.0000	0.0840	0.0667	0.0321	0.7643	0.1436
Adapt this text for a teenage audience. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	11	0.5868	0.0000	0.0000	0.0815	0.0780	0.0232	0.6970	0.0923
Adapt this text for a teenage audience. The final text must have the format <<format>>. revise · single-turn	9	0.7901	0.0000	0.0000	0.0918	0.0578	0.0340	0.6764	0.0705
Adapt this text for a teenage audience. The final text must have the topic <<topic>>. revise · single-turn	14	0.6020	0.0000	0.0000	0.0814	0.0640	0.0253	0.6562	0.1132
Adapt this text for social media sharing. revise · single & two-turn	52	0.3678	0.0000	0.0000	0.0766	0.0891	0.0234	0.7845	0.1244
Adapt this text for social media sharing. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	14	0.4592	0.0000	0.0000	0.0791	0.0774	0.0187	0.7022	0.1296
Adapt this text for social media sharing. The final text must have the format <<format>>. revise · single-turn	11	0.3884	0.0000	0.0000	0.0726	0.0810	0.0208	0.7383	0.1500
Adapt this text for social media sharing. The final text must have the topic <<topic>>. revise · single-turn	13	0.2722	0.0000	0.0000	0.0756	0.0933	0.0296	0.7792	0.1240
Change the tone of this text to be more academic and scholarly. revise · single & two-turn	69	0.4489	0.0000	0.0000	0.0811	0.0877	0.0240	0.7683	0.1411
Change the tone of this text to be more academic and scholarly. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	14	0.3214	0.0000	0.0000	0.0748	0.0852	0.0215	0.7816	0.1009
Change the tone of this text to be more academic and scholarly. The final text must have the format <<format>>. revise · single-turn	12	0.4444	0.0000	0.0000	0.0782	0.0879	0.0260	0.7366	0.0965

14Entropy (Llama-3.2-3B)AUROC 0.4751 · TPR@1% 0.0072⌄

Select a split · Entropy (Llama-3.2-3B)

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Adapt this text for a 5th-grade reading level. revise · single & two-turn	48	0.5972	0.0000	0.0000	2.3848	2.3015	0.2775	0.7738	0.1348
Adapt this text for a 5th-grade reading level. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	8	0.7500	0.0000	0.0000	2.4663	2.2239	0.2770	0.6715	0.1082
Adapt this text for a 5th-grade reading level. The final text must have the topic <<topic>>. revise · single-turn	13	0.7870	0.0000	0.0000	2.6374	2.2995	0.3625	0.7871	0.1714
Adapt this text for a lay audience with no technical background. revise · single & two-turn	54	0.6211	0.0000	0.0000	2.4548	2.2935	0.2953	0.7643	0.1436
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	12	0.5972	0.0000	0.0000	2.2477	2.0734	0.3682	0.7114	0.1210
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>>. revise · single-turn	13	0.6982	0.0000	0.0000	2.4403	2.1952	0.3069	0.7209	0.1117
Adapt this text for a teenage audience. revise · single & two-turn	63	0.5130	0.0000	0.0000	2.3988	2.4070	0.2489	0.7319	0.1237
Adapt this text for a teenage audience. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	11	0.4793	0.0000	0.0000	2.3737	2.3387	0.1946	0.6970	0.0923
Adapt this text for a teenage audience. The final text must have the format <<format>>. revise · single-turn	9	0.6790	0.0000	0.0000	2.4768	2.2401	0.2427	0.6764	0.0705
Adapt this text for a teenage audience. The final text must have the topic <<topic>>. revise · single-turn	14	0.5612	0.0000	0.0000	2.3554	2.2881	0.2559	0.6562	0.1132
Adapt this text for an elderly adult with limited knowledge of modern day affairs. revise · single & two-turn	62	0.5135	0.0000	0.0000	2.3819	2.3488	0.2261	0.7264	0.1338
Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>>. revise · single-turn	9	0.6296	0.0000	0.0000	2.2634	2.1244	0.2076	0.7067	0.1031

15FastDetectGPT (Llama-3.2-3B)AUROC 0.4595 · TPR@1% 0.0250⌄

Select a split · FastDetectGPT (Llama-3.2-3B)

Topic × formathuman · AI · |AI − human| gridsshow ▾

Human texts · mean score per topic and formatOpen full size ↗

AI texts · mean score, same colour scale as the human gridOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Generator × prompt categoryAI · |AI − human| gridsshow ▾

AI texts · mean score per generator config and prompt categoryOpen full size ↗

Mean unsigned |AI − human| score per pairOpen full size ↗

Specific prompt361 prompts · hardest firstshow ▾

One row per distinct first message (a second turn, where there is one, is ignored), hardest first: ranked by TPR at the detector's corpus-wide 1% FPR threshold. Score columns are the detector's mean over the prompt's rows, |AI − human| is the mean per-pair score movement, and distance columns are the mean distance between each source and its rewrite. Only the 12 hardest of 361 are shown; the full ranking ↗ is a CSV beside this card.

Prompt (first message)	Rows	AUROC	TPR @ 1% FPR	TPR @ 0.1% FPR	Human score	AI score	|AI − human|	jaccard_1	cosdist
Adapt this text for a 5th-grade reading level. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	8	0.3125	0.0000	0.0000	0.5858	1.2516	0.9581	0.6715	0.1082
Adapt this text for a 5th-grade reading level. The final text must have the format <<format>>. revise · single-turn	14	0.2398	0.0000	0.0000	0.0825	1.1725	1.2961	0.7458	0.1626
Adapt this text for a 5th-grade reading level. The final text must have the topic <<topic>>. revise · single-turn	13	0.2189	0.0000	0.0000	-0.6856	0.6801	1.6914	0.7871	0.1714
Adapt this text for a lay audience with no technical background. revise · single & two-turn	54	0.3800	0.0000	0.0000	-0.1622	0.3457	1.1083	0.7643	0.1436
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	12	0.2361	0.0000	0.0000	-0.6574	0.4367	1.5258	0.7114	0.1210
Adapt this text for a lay audience with no technical background. The final text must have the format <<format>>. revise · single-turn	13	0.3550	0.0000	0.0000	-0.3292	0.2833	0.9724	0.7209	0.1117
Adapt this text for a lay audience with no technical background. The final text must have the topic <<topic>>. revise · single-turn	16	0.3594	0.0000	0.0000	-0.1838	0.3628	1.3420	0.6803	0.1203
Adapt this text for a teenage audience. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	11	0.3223	0.0000	0.0000	-0.5564	-0.0234	0.8115	0.6970	0.0923
Adapt this text for a teenage audience. The final text must have the format <<format>>. revise · single-turn	9	0.2222	0.0000	0.0000	-0.4784	0.6235	1.5924	0.6764	0.0705
Adapt this text for a teenage audience. The final text must have the topic <<topic>>. revise · single-turn	14	0.3316	0.0000	0.0000	-0.2538	0.7193	1.2925	0.6562	0.1132
Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>> and topic <<topic>>. revise · single-turn	13	0.3314	0.0000	0.0000	-0.2053	0.5695	1.5107	0.7414	0.1845
Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>>. revise · single-turn	9	0.1235	0.0000	0.0000	-0.9846	0.4528	1.7225	0.7067	0.1031

03

How far do the rewrites move?

Every distance measure between each source and its rewrite: overall, then with every prompt subset and every generator config overlaid on one chart each.

jaccard_1overall · prompt subsets · generator configs⌄

OverallOpen full size ↗

One envelope per prompt subsetOpen full size ↗

One envelope per generator configOpen full size ↗

jaccard_2overall · prompt subsets · generator configs⌄

OverallOpen full size ↗

One envelope per prompt subsetOpen full size ↗

One envelope per generator configOpen full size ↗

levenshteinoverall · prompt subsets · generator configs⌄

OverallOpen full size ↗

One envelope per prompt subsetOpen full size ↗

One envelope per generator configOpen full size ↗

softngramoverall · prompt subsets · generator configs⌄

OverallOpen full size ↗

One envelope per prompt subsetOpen full size ↗

One envelope per generator configOpen full size ↗

cosdistoverall · prompt subsets · generator configs⌄

OverallOpen full size ↗

One envelope per prompt subsetOpen full size ↗

One envelope per generator configOpen full size ↗

bertscoreoverall · prompt subsets · generator configs⌄

OverallOpen full size ↗

One envelope per prompt subsetOpen full size ↗

One envelope per generator configOpen full size ↗

bertscore_precisionoverall · prompt subsets · generator configs⌄

OverallOpen full size ↗

One envelope per prompt subsetOpen full size ↗

One envelope per generator configOpen full size ↗

bertscore_recalloverall · prompt subsets · generator configs⌄

OverallOpen full size ↗

One envelope per prompt subsetOpen full size ↗

One envelope per generator configOpen full size ↗

moverscoreoverall · prompt subsets · generator configs⌄

OverallOpen full size ↗

One envelope per prompt subsetOpen full size ↗

One envelope per generator configOpen full size ↗

rerankeroverall · prompt subsets · generator configs⌄

OverallOpen full size ↗

One envelope per prompt subsetOpen full size ↗

One envelope per generator configOpen full size ↗

04

Appendix

Reference material: how the dataset was built and evaluated, its schema and loading code, univariate statistics, and cross-statistic correlations.

How the dataset was built10 generator configs · 4 prompt subsets · evaluation protocol⌄

Every row pairs original (Human) and final_response (AI): a human-written source and the text a generator model produced from it. Inclusion filtering happens while the dataset is built, so this report evaluates the published corpus exactly as it is.

Generator configs

Generator model	Temp	Sampling	Rows	Share
claude-opus-4-5-20251101	—	type disabled	1,621	10.3%
claude-opus-5	—	type disabled	1,592	10.1%
gemini-3.1-pro-preview	—	thinking_level low	1,616	10.3%
gemini-3.8-flash	—	thinking_level low	1,607	10.2%
gpt-5.4	—	reasoning_effort none	1,594	10.2%
gpt-5.6-sol	—	reasoning_effort none	1,604	10.2%
x-ai/grok-4.3	—	enabled False · max_completion_tokens 16000	1,446	9.2%
moonshotai/kimi-k3	—	enabled False · max_completion_tokens 16000	1,511	9.6%
minimax/minimax-m3	—	enabled False · max_completion_tokens 16000	1,512	9.6%
qwen/qwen3.7-max	—	enabled False · max_completion_tokens 16000	1,592	10.1%
Total			15,695	100%

Prompt subsets

Instructions are grouped by their first message. Counts cover all 15,695 rows.

direct_reference4,030 rows

single-turn · 60 instructions

Show the 60 instructions ▾

• 83× Write a new piece of text that uses this text as a citation or reference.
• 81× Write a prequel to this text, describing the events, thoughts, or historical context that immediately led up to this piece being written, making sure to use similar language and wording as the author, as if it was written by the author. The final text must have the topic <<topic>>.
• 79× Write a detailed rebuttal or critique challenging the core ideas, themes, or facts presented in this text, using the same language and roughly matching the intellectual level of the text. The final text must have the topic <<topic>>.
• 78× Take a subpoint within this text and write a full new text about it, using the language and wording from the text as appropriate. The final text must have the format <<format>>.
• 78× Write an alternate version of this text, it should follow the style and format of the original text.
• 76× Take inspiration from the style, language, and content of this text, and write a new piece of text on a topic of your choice. However, you should make sure there are slight, subtle grammatical mistakes, to make it sound like it was written by a human. The final text must have the topic <<topic>>.
• 76× Write a piece of text in the same style and language as this text, but is directly contrasting to the themes, ideas, and/or details in this text. The final text must have the topic <<topic>>.
• 75× Write a followup to this piece of text, in the same style and language as the text. The final text must have the format <<format>> and topic <<topic>>.
• 75× Write a new piece of text that is on a completely different topic, but reuses verbatim as many phrases and sentences from this text as possible.
• 74× Take a subpoint within this text and write a full new text about it, using the language and wording from the text as appropriate.
• 74× Write a different piece of text on the same topic. The final text must have the format <<format>> and topic <<topic>>.
• 74× Write a followup to this piece of text, in the same style and language as the text. The final text must have the topic <<topic>>.
• 74× Write another piece of text that resembles the style and language of this text.
• 73× Take inspiration from the style, language, and content of this text, and write a new piece of text on a topic of your choice. However, you should write the new text as if it was a rough draft being typed out on a phone.
• 73× Write a prequel to this text, describing the events, thoughts, or historical context that immediately led up to this piece being written, making sure to use similar language and wording as the author, as if it was written by the author.
• 73× Write another piece of text that is of a similar theme or background as this text. The final text must have the format <<format>> and topic <<topic>>.
• 72× Write a different piece of text on the same topic. The final text must have the format <<format>>.
• 72× Write another piece of text that is of a similar theme or background as this text. The final text must have the format <<format>>.
• 71× Take inspiration from the style, language, and content of this text, and write a new piece of text on a topic of your choice. However, you should use varied sentence lengths, and remove all common AI buzzwords such as 'tapestry', 'delve', or 'testament', to ensure that it sounds like it was written by a human. The final text must have the topic <<topic>>.
• 71× Write a detailed rebuttal or critique challenging the core ideas, themes, or facts presented in this text, using the same language and roughly matching the intellectual level of the text. The final text must have the format <<format>>.
• 71× Write a new piece of text that uses this text as a citation or reference. The final text must have the topic <<topic>>.
• 70× Take inspiration from the style, language, and content of this text, and write a new piece of text on a topic of your choice. However, you should write the new text as if it was a rough draft being typed out on a phone. The final text must have the format <<format>> and topic <<topic>>.
• 70× Write a piece of text with a different topic that the author of this text might write about. The final text must have the topic <<topic>>.
• 70× Write another piece of text that resembles the style and language of this text. The final text must have the topic <<topic>>.
• 69× Take a subpoint within this text and write a full new text about it, using the language and wording from the text as appropriate. The final text must have the format <<format>> and topic <<topic>>.
• 69× Take inspiration from the style, language, and content of this text, and write a new piece of text on a topic of your choice. However, you should make sure there are slight, subtle grammatical mistakes, to make it sound like it was written by a human.
• 69× Write a new piece of text that uses this text as a citation or reference. The final text must have the format <<format>> and topic <<topic>>.
• 69× Write a piece of text with a different topic that the author of this text might write about. The final text must have the format <<format>>.
• 68× Take inspiration from the style, language, and content of this text, and write a new piece of text on a topic of your choice. However, you should make sure there are slight, subtle grammatical mistakes, to make it sound like it was written by a human. The final text must have the format <<format>> and topic <<topic>>.
• 68× Write a followup to this piece of text, in the same style and language as the text.
• 67× Take inspiration from the style, language, and content of this text, and write a new piece of text on a topic of your choice. However, you should use varied sentence lengths, and remove all common AI buzzwords such as 'tapestry', 'delve', or 'testament', to ensure that it sounds like it was written by a human. The final text must have the format <<format>> and topic <<topic>>.
• 67× Take inspiration from the style, language, and content of this text, and write a new piece of text on a topic of your choice. However, you should use varied sentence lengths, and remove all common AI buzzwords such as 'tapestry', 'delve', or 'testament', to ensure that it sounds like it was written by a human. The final text must have the format <<format>>.
• 67× Write a piece of text in the same style and language as this text, but is directly contrasting to the themes, ideas, and/or details in this text. The final text must have the format <<format>> and topic <<topic>>.
• 67× Write a piece of text with a different topic that the author of this text might write about. The final text must have the format <<format>> and topic <<topic>>.
• 66× Take a subpoint within this text and write a full new text about it, using the language and wording from the text as appropriate. The final text must have the topic <<topic>>.
• 66× Write a piece of text in the same style and language as this text, but is directly contrasting to the themes, ideas, and/or details in this text.
• 66× Write a piece of text with a different topic that the author of this text might write about.
• 66× Write a prequel to this text, describing the events, thoughts, or historical context that immediately led up to this piece being written, making sure to use similar language and wording as the author, as if it was written by the author. The final text must have the format <<format>>.
• 66× Write another piece of text that is of a similar theme or background as this text.
• 66× Write another piece of text that resembles the style and language of this text. The final text must have the format <<format>>.
• 65× Write a new piece of text that uses this text as a citation or reference. The final text must have the format <<format>>.
• 64× Write a prequel to this text, describing the events, thoughts, or historical context that immediately led up to this piece being written, making sure to use similar language and wording as the author, as if it was written by the author. The final text must have the format <<format>> and topic <<topic>>.
• 63× Write a detailed rebuttal or critique challenging the core ideas, themes, or facts presented in this text, using the same language and roughly matching the intellectual level of the text.
• 62× Write an alternate version of this text, it should follow the style and format of the original text. The final text must have the format <<format>> and topic <<topic>>.
• 61× Write a detailed rebuttal or critique challenging the core ideas, themes, or facts presented in this text, using the same language and roughly matching the intellectual level of the text. The final text must have the format <<format>> and topic <<topic>>.
• 61× Write a different piece of text on the same topic. The final text must have the topic <<topic>>.
• 60× Take inspiration from the style, language, and content of this text, and write a new piece of text on a topic of your choice. However, you should write the new text as if it was a rough draft being typed out on a phone. The final text must have the format <<format>>.
• 60× Write a different piece of text on the same topic.
• 59× Write another piece of text that resembles the style and language of this text. The final text must have the format <<format>> and topic <<topic>>.
• 58× Take inspiration from the style, language, and content of this text, and write a new piece of text on a topic of your choice. However, you should make sure there are slight, subtle grammatical mistakes, to make it sound like it was written by a human. The final text must have the format <<format>>.
• 58× Write an alternate version of this text, it should follow the style and format of the original text. The final text must have the format <<format>>.
• 57× Take inspiration from the style, language, and content of this text, and write a new piece of text on a topic of your choice. However, you should use varied sentence lengths, and remove all common AI buzzwords such as 'tapestry', 'delve', or 'testament', to ensure that it sounds like it was written by a human.
• 57× Write a new piece of text that is on a completely different topic, but reuses verbatim as many phrases and sentences from this text as possible. The final text must have the format <<format>>.
• 57× Write a piece of text in the same style and language as this text, but is directly contrasting to the themes, ideas, and/or details in this text. The final text must have the format <<format>>.
• 57× Write an alternate version of this text, it should follow the style and format of the original text. The final text must have the topic <<topic>>.
• 56× Write a new piece of text that is on a completely different topic, but reuses verbatim as many phrases and sentences from this text as possible. The final text must have the format <<format>> and topic <<topic>>.
• 56× Write another piece of text that is of a similar theme or background as this text. The final text must have the topic <<topic>>.
• 55× Write a new piece of text that is on a completely different topic, but reuses verbatim as many phrases and sentences from this text as possible. The final text must have the topic <<topic>>.
• 53× Write a followup to this piece of text, in the same style and language as the text. The final text must have the format <<format>>.
• 52× Take inspiration from the style, language, and content of this text, and write a new piece of text on a topic of your choice. However, you should write the new text as if it was a rough draft being typed out on a phone. The final text must have the topic <<topic>>.

indirect_reference4,001 rows

two-turn · 25 instructions

Show the 25 instructions ▾

• 218× Translate the given text to Chinese. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 216× Translate the given text to French. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 216× Translate the given text to Spanish. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 215× Translate the given text to German. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 199× Formulate a set of strict, absolute constraints regarding format, word count, vocabulary restrictions, and syntax rules that, if followed perfectly by an AI, would result in generating a text nearly identical to this one. Note that your response should be formatted as a prompt that would be sent to an AI requesting it to generate something, not just a list of rules. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 193× Translate the given text to Hindi. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 170× Convert this text into a series of logical propositions or syllogisms that represent the core argument or narrative progression, stripped of all rhetorical flair. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 170× This piece of text was cleverely generated by an LLM with a human supervising it so that the text seems authentic. Reverse engineer the process and come up with a prompt that is most probable to be the one used to generate the text. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 169× Create a prompt that might cause an LLM to generate an output resembling this text. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 167× Summarize this text, attempting to preserve as much of the original language of the text as possible. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 165× Write a 'few-shot' prompt using minimal, fragmented excerpts from this text as examples. The prompt should be designed to teach an LLM the pattern needed to generate the rest of the text from scratch. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 159× Take this text, but extract only the most meaningful sentences out of it to create a new text that is a series of disjoint sentences representing the original text. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 156× Replace every verb in this text with the tag [VERB], keeping all nouns, adjectives, articles and punctuation exactly as they appear. Reproduce the document in full, in its original order and paragraph structure. Do not output a legend or a list of the tags you used. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 153× Take this text and replace one word in every four with a single underscore (use one underscore for each removed word, never several in a row). This way, it will be appropriate for a FIM task. You may decide which words to replace. Reproduce the rest of the document in full, in its original order and paragraph structure. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 150× Create a prompt that would allow an LLM to generate this text as accurately/verbatim as possible, without the prompt containing any of the actual text. That is, you must use descriptors to accurately describe the text to generate. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 146× Provide only the first and last word of every sentence in this text, replacing the middle of each sentence with an ellipsis '...'. Maintain the original paragraph structure. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 145× Create the shortest possible prompt that can be sent to an LLM to reproduce this text. The prompt may include verbatim pieces of the original text, however, it must be short and accurate, working as an effective compression of the original text. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 139× Write a detailed descriptor for how this text is stylistically differentiated from other texts, including grammatical choices and word choices not present in other texts of this kind. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 132× Envision the scenario in which the author wrote this text. Describe that scenario in exhaustive detail. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 129× Create a detailed dictionary that fully describes all the meaningful phrases and words this author utilizes to create this text, and how the author utilizes those words and phrases, and why. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 129× Write a detailed descriptor for the manner in which the author of this uses language, and things such as their word choice. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 127× Write a detailed descriptor for the style of writing of this text, including its qualitative and quantitative properties. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 121× Reformat this text into a sensible, structured JSON object. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 111× Write a detailed descriptor for the (imaginary) personality of the author who wrote this text, including their livelihood and day to day activities, as well as how that influences their writing. Do not output anything besides what you were requested to write, and do not output any extra commentary.
• 106× Translate the entirety of this text into a sequence of emojis that captures the literal meaning, tone, and narrative progression of the writing. Do not use any letters or words. Do not output anything besides what you were requested to write, and do not output any extra commentary.

revise4,068 rows

2,001 single · 2,067 two-turn · 172 instructions

Show the 172 instructions ▾

• 77× Change the tone of this text to be more objective and unbiased.
• 74× Restructure this text to eliminate awkward phrasing and improve the overall rhythm.
• 73× Clarify this text by making the cause-and-effect relationships more obvious.
• 73× Clarify this text by presenting the same information from a fresh angle.
• 72× Clarify this text by eliminating any ambiguous or vague language.
• 71× Change the tone of this text to be more suited for a business audience.
• 69× Change the tone of this text to be more academic and scholarly.
• 69× Change the tone of this text to be more urgent and compelling.
• 68× Restructure this text to arrange the points in order of importance.
• 67× Expand this text by adding vivid imagery and sensory details to bring it to life.
• 67× Restructure this text to improve the transitions and connections between ideas.
• 66× Expand this text by adding storytelling elements and dialogue to increase engagement.
• 65× Change the tone of this text to be more inspiring and motivational.
• 65× Change the tone of this text to be more lighthearted and humorous.
• 64× Change the tone of this text to be more persuasive and convincing.
• 64× Clarify this text by paraphrasing it with different vocabulary and varied sentence structures.
• 63× Adapt this text for a teenage audience.
• 63× Restructure this text to improve the logical flow and argument progression.
• 62× Adapt this text for an elderly adult with limited knowledge of modern day affairs.
• 60× Change the tone of this text to be more empathetic and understanding.
• 60× Edit this text to eliminate redundant and filler words.
• 60× Expand this text by adding backstory and context to enrich understanding.
• 60× Restructure this text to group related ideas and include clear section breaks.
• 59× Restructure this text to read more naturally, as if written by a native English speaker.
• 58× Edit this text to trim the length without losing important information or tone.
• 58× Expand this text by adding real-world applications and specific details.
• 57× Change the tone of this text to be more enthusiastic and energetic.
• 57× Change the tone of this text to be more professional and authoritative.
• 57× Clarify this text by making the argument and main points stronger and easier to understand.
• 55× Change the tone of this text to be more analytical and logical.
• 55× Clarify this text by improving the phrasing while preserving its original intent.
• 54× Adapt this text for a lay audience with no technical background.
• 54× Expand this text by adding concrete examples to illustrate the main points.
• 53× Change the tone of this text to be more direct and bold.
• 52× Adapt this text for social media sharing.
• 51× Clarify this text by defining any terms that might be unclear to the reader.
• 50× Edit this text to be more concise and powerful.
• 49× Expand this text by adding supporting information to elaborate on the key points.
• 48× Adapt this text for a 5th-grade reading level.
• 48× Change the tone of this text to be more conversational and approachable.
• 47× Change the tone of this text to be more relaxed and friendly.
• 45× Restructure this text to ensure a clear introduction, body, and compelling conclusion.
• 43× Change the tone of this text to be more polite and diplomatic.
• 20× Clarify this text by presenting the same information from a fresh angle. The final text must have the format <<format>> and topic <<topic>>.
• 19× Change the tone of this text to be more direct and bold. The final text must have the format <<format>> and topic <<topic>>.
• 19× Change the tone of this text to be more relaxed and friendly. The final text must have the format <<format>>.
• 18× Change the tone of this text to be more enthusiastic and energetic. The final text must have the format <<format>>.
• 18× Change the tone of this text to be more objective and unbiased. The final text must have the format <<format>>.
• 18× Edit this text to trim the length without losing important information or tone. The final text must have the format <<format>> and topic <<topic>>.
• 17× Change the tone of this text to be more polite and diplomatic. The final text must have the format <<format>>.
• 17× Change the tone of this text to be more suited for a business audience. The final text must have the topic <<topic>>.
• 17× Expand this text by adding concrete examples to illustrate the main points. The final text must have the format <<format>>.
• 17× Restructure this text to read more naturally, as if written by a native English speaker. The final text must have the topic <<topic>>.
• 16× Adapt this text for a lay audience with no technical background. The final text must have the topic <<topic>>.
• 16× Change the tone of this text to be more direct and bold. The final text must have the topic <<topic>>.
• 16× Change the tone of this text to be more empathetic and understanding. The final text must have the format <<format>> and topic <<topic>>.
• 16× Change the tone of this text to be more inspiring and motivational. The final text must have the format <<format>>.
• 16× Restructure this text to group related ideas and include clear section breaks. The final text must have the topic <<topic>>.
• 15× Change the tone of this text to be more enthusiastic and energetic. The final text must have the format <<format>> and topic <<topic>>.
• 15× Clarify this text by making the argument and main points stronger and easier to understand. The final text must have the format <<format>>.
• 15× Clarify this text by paraphrasing it with different vocabulary and varied sentence structures. The final text must have the format <<format>> and topic <<topic>>.
• 15× Expand this text by adding storytelling elements and dialogue to increase engagement. The final text must have the format <<format>>.
• 15× Restructure this text to group related ideas and include clear section breaks. The final text must have the format <<format>> and topic <<topic>>.
• 15× Restructure this text to improve the logical flow and argument progression. The final text must have the topic <<topic>>.
• 14× Adapt this text for a 5th-grade reading level. The final text must have the format <<format>>.
• 14× Adapt this text for a teenage audience. The final text must have the topic <<topic>>.
• 14× Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the topic <<topic>>.
• 14× Adapt this text for social media sharing. The final text must have the format <<format>> and topic <<topic>>.
• 14× Change the tone of this text to be more academic and scholarly. The final text must have the format <<format>> and topic <<topic>>.
• 14× Change the tone of this text to be more conversational and approachable. The final text must have the format <<format>>.
• 14× Change the tone of this text to be more empathetic and understanding. The final text must have the format <<format>>.
• 14× Change the tone of this text to be more lighthearted and humorous. The final text must have the format <<format>> and topic <<topic>>.
• 14× Change the tone of this text to be more suited for a business audience. The final text must have the format <<format>>.
• 14× Clarify this text by eliminating any ambiguous or vague language. The final text must have the format <<format>> and topic <<topic>>.
• 14× Edit this text to be more concise and powerful. The final text must have the topic <<topic>>.
• 14× Expand this text by adding backstory and context to enrich understanding. The final text must have the format <<format>> and topic <<topic>>.
• 14× Expand this text by adding real-world applications and specific details. The final text must have the topic <<topic>>.
• 14× Expand this text by adding supporting information to elaborate on the key points. The final text must have the format <<format>> and topic <<topic>>.
• 14× Restructure this text to arrange the points in order of importance. The final text must have the topic <<topic>>.
• 14× Restructure this text to eliminate awkward phrasing and improve the overall rhythm. The final text must have the format <<format>> and topic <<topic>>.
• 13× Adapt this text for a 5th-grade reading level. The final text must have the topic <<topic>>.
• 13× Adapt this text for a lay audience with no technical background. The final text must have the format <<format>>.
• 13× Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>> and topic <<topic>>.
• 13× Adapt this text for social media sharing. The final text must have the topic <<topic>>.
• 13× Change the tone of this text to be more conversational and approachable. The final text must have the format <<format>> and topic <<topic>>.
• 13× Clarify this text by presenting the same information from a fresh angle. The final text must have the format <<format>>.
• 13× Clarify this text by presenting the same information from a fresh angle. The final text must have the topic <<topic>>.
• 13× Restructure this text to read more naturally, as if written by a native English speaker. The final text must have the format <<format>> and topic <<topic>>.
• 12× Adapt this text for a lay audience with no technical background. The final text must have the format <<format>> and topic <<topic>>.
• 12× Change the tone of this text to be more academic and scholarly. The final text must have the format <<format>>.
• 12× Change the tone of this text to be more direct and bold. The final text must have the format <<format>>.
• 12× Change the tone of this text to be more inspiring and motivational. The final text must have the topic <<topic>>.
• 12× Change the tone of this text to be more lighthearted and humorous. The final text must have the format <<format>>.
• 12× Change the tone of this text to be more persuasive and convincing. The final text must have the format <<format>> and topic <<topic>>.
• 12× Change the tone of this text to be more persuasive and convincing. The final text must have the topic <<topic>>.
• 12× Change the tone of this text to be more relaxed and friendly. The final text must have the format <<format>> and topic <<topic>>.
• 12× Change the tone of this text to be more urgent and compelling. The final text must have the topic <<topic>>.
• 12× Clarify this text by defining any terms that might be unclear to the reader. The final text must have the format <<format>>.
• 12× Clarify this text by defining any terms that might be unclear to the reader. The final text must have the topic <<topic>>.
• 12× Clarify this text by improving the phrasing while preserving its original intent. The final text must have the format <<format>>.
• 12× Clarify this text by making the cause-and-effect relationships more obvious. The final text must have the format <<format>> and topic <<topic>>.
• 12× Edit this text to eliminate redundant and filler words. The final text must have the format <<format>> and topic <<topic>>.
• 12× Expand this text by adding real-world applications and specific details. The final text must have the format <<format>> and topic <<topic>>.
• 12× Expand this text by adding real-world applications and specific details. The final text must have the format <<format>>.
• 12× Expand this text by adding supporting information to elaborate on the key points. The final text must have the topic <<topic>>.
• 12× Expand this text by adding vivid imagery and sensory details to bring it to life. The final text must have the topic <<topic>>.
• 12× Restructure this text to eliminate awkward phrasing and improve the overall rhythm. The final text must have the format <<format>>.
• 12× Restructure this text to read more naturally, as if written by a native English speaker. The final text must have the format <<format>>.
• 11× Adapt this text for a teenage audience. The final text must have the format <<format>> and topic <<topic>>.
• 11× Adapt this text for social media sharing. The final text must have the format <<format>>.
• 11× Change the tone of this text to be more analytical and logical. The final text must have the format <<format>> and topic <<topic>>.
• 11× Change the tone of this text to be more analytical and logical. The final text must have the topic <<topic>>.
• 11× Change the tone of this text to be more conversational and approachable. The final text must have the topic <<topic>>.
• 11× Change the tone of this text to be more objective and unbiased. The final text must have the format <<format>> and topic <<topic>>.
• 11× Change the tone of this text to be more persuasive and convincing. The final text must have the format <<format>>.
• 11× Change the tone of this text to be more polite and diplomatic. The final text must have the format <<format>> and topic <<topic>>.
• 11× Edit this text to eliminate redundant and filler words. The final text must have the format <<format>>.
• 11× Edit this text to trim the length without losing important information or tone. The final text must have the topic <<topic>>.
• 11× Expand this text by adding vivid imagery and sensory details to bring it to life. The final text must have the format <<format>> and topic <<topic>>.
• 11× Expand this text by adding vivid imagery and sensory details to bring it to life. The final text must have the format <<format>>.
• 11× Restructure this text to group related ideas and include clear section breaks. The final text must have the format <<format>>.
• 11× Restructure this text to improve the logical flow and argument progression. The final text must have the format <<format>> and topic <<topic>>.
• 11× Restructure this text to improve the transitions and connections between ideas. The final text must have the format <<format>>.
• 10× Change the tone of this text to be more inspiring and motivational. The final text must have the format <<format>> and topic <<topic>>.
• 10× Change the tone of this text to be more polite and diplomatic. The final text must have the topic <<topic>>.
• 10× Change the tone of this text to be more professional and authoritative. The final text must have the topic <<topic>>.
• 10× Clarify this text by eliminating any ambiguous or vague language. The final text must have the topic <<topic>>.
• 10× Clarify this text by making the argument and main points stronger and easier to understand. The final text must have the format <<format>> and topic <<topic>>.
• 10× Clarify this text by paraphrasing it with different vocabulary and varied sentence structures. The final text must have the format <<format>>.
• 10× Clarify this text by paraphrasing it with different vocabulary and varied sentence structures. The final text must have the topic <<topic>>.
• 10× Edit this text to be more concise and powerful. The final text must have the format <<format>> and topic <<topic>>.
• 10× Edit this text to eliminate redundant and filler words. The final text must have the topic <<topic>>.
• 10× Restructure this text to arrange the points in order of importance. The final text must have the format <<format>> and topic <<topic>>.
• 10× Restructure this text to ensure a clear introduction, body, and compelling conclusion. The final text must have the format <<format>> and topic <<topic>>.
• 10× Restructure this text to ensure a clear introduction, body, and compelling conclusion. The final text must have the topic <<topic>>.
• 9× Adapt this text for a teenage audience. The final text must have the format <<format>>.
• 9× Adapt this text for an elderly adult with limited knowledge of modern day affairs. The final text must have the format <<format>>.
• 9× Change the tone of this text to be more analytical and logical. The final text must have the format <<format>>.
• 9× Change the tone of this text to be more lighthearted and humorous. The final text must have the topic <<topic>>.
• 9× Change the tone of this text to be more objective and unbiased. The final text must have the topic <<topic>>.
• 9× Change the tone of this text to be more professional and authoritative. The final text must have the format <<format>> and topic <<topic>>.
• 9× Change the tone of this text to be more urgent and compelling. The final text must have the format <<format>> and topic <<topic>>.
• 9× Clarify this text by eliminating any ambiguous or vague language. The final text must have the format <<format>>.
• 9× Clarify this text by making the argument and main points stronger and easier to understand. The final text must have the topic <<topic>>.
• 9× Clarify this text by making the cause-and-effect relationships more obvious. The final text must have the format <<format>>.
• 9× Expand this text by adding backstory and context to enrich understanding. The final text must have the topic <<topic>>.
• 9× Expand this text by adding concrete examples to illustrate the main points. The final text must have the topic <<topic>>.
• 9× Restructure this text to arrange the points in order of importance. The final text must have the format <<format>>.
• 9× Restructure this text to ensure a clear introduction, body, and compelling conclusion. The final text must have the format <<format>>.
• 9× Restructure this text to improve the transitions and connections between ideas. The final text must have the format <<format>> and topic <<topic>>.
• 8× Adapt this text for a 5th-grade reading level. The final text must have the format <<format>> and topic <<topic>>.
• 8× Change the tone of this text to be more academic and scholarly. The final text must have the topic <<topic>>.
• 8× Change the tone of this text to be more relaxed and friendly. The final text must have the topic <<topic>>.
• 8× Change the tone of this text to be more urgent and compelling. The final text must have the format <<format>>.
• 8× Clarify this text by improving the phrasing while preserving its original intent. The final text must have the format <<format>> and topic <<topic>>.
• 8× Expand this text by adding backstory and context to enrich understanding. The final text must have the format <<format>>.
• 8× Expand this text by adding storytelling elements and dialogue to increase engagement. The final text must have the format <<format>> and topic <<topic>>.
• 8× Expand this text by adding storytelling elements and dialogue to increase engagement. The final text must have the topic <<topic>>.
• 8× Expand this text by adding supporting information to elaborate on the key points. The final text must have the format <<format>>.
• 8× Restructure this text to improve the logical flow and argument progression. The final text must have the format <<format>>.
• 8× Restructure this text to improve the transitions and connections between ideas. The final text must have the topic <<topic>>.
• 7× Change the tone of this text to be more suited for a business audience. The final text must have the format <<format>> and topic <<topic>>.
• 7× Restructure this text to eliminate awkward phrasing and improve the overall rhythm. The final text must have the topic <<topic>>.
• 6× Change the tone of this text to be more empathetic and understanding. The final text must have the topic <<topic>>.
• 6× Clarify this text by improving the phrasing while preserving its original intent. The final text must have the topic <<topic>>.
• 6× Clarify this text by making the cause-and-effect relationships more obvious. The final text must have the topic <<topic>>.
• 6× Edit this text to be more concise and powerful. The final text must have the format <<format>>.
• 5× Change the tone of this text to be more professional and authoritative. The final text must have the format <<format>>.
• 5× Clarify this text by defining any terms that might be unclear to the reader. The final text must have the format <<format>> and topic <<topic>>.
• 5× Expand this text by adding concrete examples to illustrate the main points. The final text must have the format <<format>> and topic <<topic>>.
• 4× Change the tone of this text to be more enthusiastic and energetic. The final text must have the topic <<topic>>.
• 3× Edit this text to trim the length without losing important information or tone. The final text must have the format <<format>>.

rewrite3,596 rows

1,701 single · 1,895 two-turn · 104 instructions

Show the 104 instructions ▾

• 104× Edit this text by upgrading the verbs to be more dynamic and action-oriented, leaving the subjects, objects, and overall length completely unchanged.
• 104× Rewrite the weakest section in this text, but do not make any modifications to any other section.
• 102× Modify this text so that it is as short as possible. At least a quarter of the original text must be repeated verbatim, and at most half of it; the remainder must be new prose.
• 101× Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound.
• 101× Edit this text by replacing all existing transition words and phrases (e.g., 'however', 'therefore', 'in addition') with different, synonymous transitions, making no other changes.
• 100× Edit this text by completely rewriting the topic sentence of every paragraph, while preserving the supporting sentences in those paragraphs verbatim.
• 97× Edit this text by replacing exactly one to three sentences, without changing anything else.
• 93× Edit this text by removing one section from it (the least necessary), you may only make minimal modifications to adjacent sections for better flow.
• 93× Edit this text by replacing all instances of passive voice with active voice, without altering any sentences that are already in the active voice.
• 91× Edit this text by inserting a single, highly engaging introductory sentence at the very beginning, without modifying any of the original text that follows.
• 91× Rewrite only the opening hook and the final concluding sentence of this text, leaving the entire middle portion exactly as it was originally written.
• 89× Edit this text by removing exactly one to three sentences from it (the least necessary), you may only make minimal modifications to adjacent sentences for better flow.
• 89× Edit this text by removing five to ten percent of the words from it (the least necessary), without changing anything else.
• 87× Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched.
• 87× Rewrite exactly one section of this text so that its wording is entirely new, and leave every other section untouched. The rewritten section must share no sentence with the original.
• 86× Edit this text by replacing generic adjectives with highly specific, evocative ones, modifying no more than ten percent of the total words.
• 86× Modify this text to be as long as possible. At least half of the original text must be repeated verbatim, and at most three quarters of it; the remainder must be new prose.
• 83× Edit this text by inserting exactly one to three sentences (in appropriate locations), you may only make minimal modifications to transitions in adjacent sentences for better flow.
• 82× Improve this text by rewriting it. At least half of the original text must be repeated verbatim, and at most three quarters of it; the remainder must be new prose.
• 81× Edit this text by inserting a singular new section (in an appropriate locations), you may only make minimal modifications to transitions in adjacent sections for better flow.
• 81× Edit this text by replacing the final sentence with a more open-ended, thought-provoking statement, without altering any preceding text.
• 81× Edit this text by splitting the three longest sentences into shorter, distinct sentences, without changing the vocabulary or structure of the rest of the text.
• 75× Edit this text by inserting twenty to thirty words (in appropriate locations), without changing anything else.
• 75× Edit this text by replacing five to ten percent of words in it, without changing anything else.
• 74× Edit this text by replacing, removing or inserting exactly one to three sentences, without changing anything else.
• 63× Edit this text by replacing, removing or inserting twenty to thirty words, without changing anything else.
• 28× Edit this text by completely rewriting the topic sentence of every paragraph, while preserving the supporting sentences in those paragraphs verbatim. The final text must have the format <<format>>.
• 26× Edit this text by completely rewriting the topic sentence of every paragraph, while preserving the supporting sentences in those paragraphs verbatim. The final text must have the format <<format>> and topic <<topic>>.
• 26× Improve this text by rewriting it. At least half of the original text must be repeated verbatim, and at most three quarters of it; the remainder must be new prose. The final text must have the format <<format>> and topic <<topic>>.
• 24× Edit this text by inserting a single, highly engaging introductory sentence at the very beginning, without modifying any of the original text that follows. The final text must have the topic <<topic>>.
• 24× Edit this text by removing one section from it (the least necessary), you may only make minimal modifications to adjacent sections for better flow. The final text must have the format <<format>>.
• 24× Improve this text by rewriting it. At least half of the original text must be repeated verbatim, and at most three quarters of it; the remainder must be new prose. The final text must have the topic <<topic>>.
• 23× Edit this text by inserting exactly one to three sentences (in appropriate locations), you may only make minimal modifications to transitions in adjacent sentences for better flow. The final text must have the format <<format>> and topic <<topic>>.
• 23× Edit this text by upgrading the verbs to be more dynamic and action-oriented, leaving the subjects, objects, and overall length completely unchanged. The final text must have the topic <<topic>>.
• 22× Edit this text by removing one section from it (the least necessary), you may only make minimal modifications to adjacent sections for better flow. The final text must have the topic <<topic>>.
• 22× Edit this text by replacing all instances of passive voice with active voice, without altering any sentences that are already in the active voice. The final text must have the topic <<topic>>.
• 21× Edit this text by inserting a single, highly engaging introductory sentence at the very beginning, without modifying any of the original text that follows. The final text must have the format <<format>>.
• 21× Edit this text by inserting exactly one to three sentences (in appropriate locations), you may only make minimal modifications to transitions in adjacent sentences for better flow. The final text must have the format <<format>>.
• 21× Edit this text by inserting twenty to thirty words (in appropriate locations), without changing anything else. The final text must have the format <<format>>.
• 21× Edit this text by replacing, removing or inserting exactly one to three sentences, without changing anything else. The final text must have the format <<format>> and topic <<topic>>.
• 21× Modify this text to be as long as possible. At least half of the original text must be repeated verbatim, and at most three quarters of it; the remainder must be new prose. The final text must have the format <<format>> and topic <<topic>>.
• 20× Edit this text by inserting a singular new section (in an appropriate locations), you may only make minimal modifications to transitions in adjacent sections for better flow. The final text must have the format <<format>> and topic <<topic>>.
• 20× Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound. The final text must have the format <<format>>.
• 20× Edit this text by removing five to ten percent of the words from it (the least necessary), without changing anything else. The final text must have the format <<format>> and topic <<topic>>.
• 20× Edit this text by removing one section from it (the least necessary), you may only make minimal modifications to adjacent sections for better flow. The final text must have the format <<format>> and topic <<topic>>.
• 20× Edit this text by replacing the final sentence with a more open-ended, thought-provoking statement, without altering any preceding text. The final text must have the topic <<topic>>.
• 20× Modify this text so that it is as short as possible. At least a quarter of the original text must be repeated verbatim, and at most half of it; the remainder must be new prose. The final text must have the format <<format>> and topic <<topic>>.
• 20× Rewrite only the opening hook and the final concluding sentence of this text, leaving the entire middle portion exactly as it was originally written. The final text must have the format <<format>> and topic <<topic>>.
• 20× Rewrite the weakest section in this text, but do not make any modifications to any other section. The final text must have the topic <<topic>>.
• 19× Edit this text by inserting a single, highly engaging introductory sentence at the very beginning, without modifying any of the original text that follows. The final text must have the format <<format>> and topic <<topic>>.
• 19× Edit this text by replacing all instances of passive voice with active voice, without altering any sentences that are already in the active voice. The final text must have the format <<format>>.
• 19× Edit this text by replacing five to ten percent of words in it, without changing anything else. The final text must have the format <<format>> and topic <<topic>>.
• 19× Modify this text so that it is as short as possible. At least a quarter of the original text must be repeated verbatim, and at most half of it; the remainder must be new prose. The final text must have the format <<format>>.
• 19× Modify this text to be as long as possible. At least half of the original text must be repeated verbatim, and at most three quarters of it; the remainder must be new prose. The final text must have the topic <<topic>>.
• 18× Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched. The final text must have the topic <<topic>>.
• 18× Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound. The final text must have the format <<format>> and topic <<topic>>.
• 18× Edit this text by splitting the three longest sentences into shorter, distinct sentences, without changing the vocabulary or structure of the rest of the text. The final text must have the format <<format>>.
• 18× Edit this text by upgrading the verbs to be more dynamic and action-oriented, leaving the subjects, objects, and overall length completely unchanged. The final text must have the format <<format>>.
• 18× Modify this text so that it is as short as possible. At least a quarter of the original text must be repeated verbatim, and at most half of it; the remainder must be new prose. The final text must have the topic <<topic>>.
• 18× Rewrite exactly one section of this text so that its wording is entirely new, and leave every other section untouched. The rewritten section must share no sentence with the original. The final text must have the topic <<topic>>.
• 18× Rewrite the weakest section in this text, but do not make any modifications to any other section. The final text must have the format <<format>> and topic <<topic>>.
• 17× Edit this text by inserting twenty to thirty words (in appropriate locations), without changing anything else. The final text must have the topic <<topic>>.
• 17× Edit this text by replacing all existing transition words and phrases (e.g., 'however', 'therefore', 'in addition') with different, synonymous transitions, making no other changes. The final text must have the format <<format>> and topic <<topic>>.
• 17× Edit this text by replacing exactly one to three sentences, without changing anything else. The final text must have the format <<format>>.
• 17× Edit this text by splitting the three longest sentences into shorter, distinct sentences, without changing the vocabulary or structure of the rest of the text. The final text must have the topic <<topic>>.
• 17× Rewrite exactly one section of this text so that its wording is entirely new, and leave every other section untouched. The rewritten section must share no sentence with the original. The final text must have the format <<format>> and topic <<topic>>.
• 16× Edit this text by inserting twenty to thirty words (in appropriate locations), without changing anything else. The final text must have the format <<format>> and topic <<topic>>.
• 16× Edit this text by replacing all instances of passive voice with active voice, without altering any sentences that are already in the active voice. The final text must have the format <<format>> and topic <<topic>>.
• 16× Edit this text by replacing exactly one to three sentences, without changing anything else. The final text must have the format <<format>> and topic <<topic>>.
• 16× Edit this text by replacing the final sentence with a more open-ended, thought-provoking statement, without altering any preceding text. The final text must have the format <<format>>.
• 16× Edit this text by splitting the three longest sentences into shorter, distinct sentences, without changing the vocabulary or structure of the rest of the text. The final text must have the format <<format>> and topic <<topic>>.
• 15× Edit this text by completely rewriting the topic sentence of every paragraph, while preserving the supporting sentences in those paragraphs verbatim. The final text must have the topic <<topic>>.
• 15× Edit this text by inserting a singular new section (in an appropriate locations), you may only make minimal modifications to transitions in adjacent sections for better flow. The final text must have the topic <<topic>>.
• 15× Edit this text by inserting exactly one to three sentences (in appropriate locations), you may only make minimal modifications to transitions in adjacent sentences for better flow. The final text must have the topic <<topic>>.
• 15× Edit this text by removing exactly one to three sentences from it (the least necessary), you may only make minimal modifications to adjacent sentences for better flow. The final text must have the format <<format>> and topic <<topic>>.
• 15× Edit this text by removing exactly one to three sentences from it (the least necessary), you may only make minimal modifications to adjacent sentences for better flow. The final text must have the format <<format>>.
• 15× Edit this text by replacing five to ten percent of words in it, without changing anything else. The final text must have the format <<format>>.
• 15× Edit this text by replacing five to ten percent of words in it, without changing anything else. The final text must have the topic <<topic>>.
• 15× Modify this text to be as long as possible. At least half of the original text must be repeated verbatim, and at most three quarters of it; the remainder must be new prose. The final text must have the format <<format>>.
• 15× Rewrite only the opening hook and the final concluding sentence of this text, leaving the entire middle portion exactly as it was originally written. The final text must have the topic <<topic>>.
• 14× Edit this text by removing five to ten percent of the words from it (the least necessary), without changing anything else. The final text must have the format <<format>>.
• 14× Edit this text by replacing the final sentence with a more open-ended, thought-provoking statement, without altering any preceding text. The final text must have the format <<format>> and topic <<topic>>.
• 14× Rewrite exactly one section of this text so that its wording is entirely new, and leave every other section untouched. The rewritten section must share no sentence with the original. The final text must have the format <<format>>.
• 14× Rewrite only the opening hook and the final concluding sentence of this text, leaving the entire middle portion exactly as it was originally written. The final text must have the format <<format>>.
• 13× Edit this text by removing five to ten percent of the words from it (the least necessary), without changing anything else. The final text must have the topic <<topic>>.
• 13× Edit this text by replacing generic adjectives with highly specific, evocative ones, modifying no more than ten percent of the total words. The final text must have the format <<format>>.
• 13× Edit this text by replacing, removing or inserting exactly one to three sentences, without changing anything else. The final text must have the format <<format>>.
• 13× Edit this text by replacing, removing or inserting twenty to thirty words, without changing anything else. The final text must have the format <<format>> and topic <<topic>>.
• 13× Edit this text by upgrading the verbs to be more dynamic and action-oriented, leaving the subjects, objects, and overall length completely unchanged. The final text must have the format <<format>> and topic <<topic>>.
• 13× Rewrite the weakest section in this text, but do not make any modifications to any other section. The final text must have the format <<format>>.
• 12× Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched. The final text must have the format <<format>>.
• 12× Edit this text by inserting a singular new section (in an appropriate locations), you may only make minimal modifications to transitions in adjacent sections for better flow. The final text must have the format <<format>>.
• 12× Edit this text by replacing exactly one to three sentences, without changing anything else. The final text must have the topic <<topic>>.
• 12× Edit this text by replacing generic adjectives with highly specific, evocative ones, modifying no more than ten percent of the total words. The final text must have the format <<format>> and topic <<topic>>.
• 12× Edit this text by replacing generic adjectives with highly specific, evocative ones, modifying no more than ten percent of the total words. The final text must have the topic <<topic>>.
• 11× Edit this text by removing exactly one to three sentences from it (the least necessary), you may only make minimal modifications to adjacent sentences for better flow. The final text must have the topic <<topic>>.
• 10× Edit this text by replacing all existing transition words and phrases (e.g., 'however', 'therefore', 'in addition') with different, synonymous transitions, making no other changes. The final text must have the format <<format>>.
• 10× Improve this text by rewriting it. At least half of the original text must be repeated verbatim, and at most three quarters of it; the remainder must be new prose. The final text must have the format <<format>>.
• 9× Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched. The final text must have the format <<format>> and topic <<topic>>.
• 9× Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound. The final text must have the topic <<topic>>.
• 8× Edit this text by replacing all existing transition words and phrases (e.g., 'however', 'therefore', 'in addition') with different, synonymous transitions, making no other changes. The final text must have the topic <<topic>>.
• 8× Edit this text by replacing, removing or inserting twenty to thirty words, without changing anything else. The final text must have the topic <<topic>>.
• 7× Edit this text by replacing, removing or inserting twenty to thirty words, without changing anything else. The final text must have the format <<format>>.
• 6× Edit this text by replacing, removing or inserting exactly one to three sentences, without changing anything else. The final text must have the topic <<topic>>.

Topics

Topic	Rows	Share
Finance & Business	1,122	7.1%
Education & Jobs	899	5.7%
Health	856	5.5%
Software Dev.	789	5.0%
Sports & Fitness	775	4.9%
Literature	771	4.9%
Entertainment	751	4.8%
Crime & Law	724	4.6%
Social Life	715	4.6%
Science & Tech.	714	4.5%
Politics	664	4.2%
Software	644	4.1%
Religion	622	4.0%
Hardware	609	3.9%
Transportation	595	3.8%
Industrial	586	3.7%
Home & Hobbies	581	3.7%
Food & Dining	577	3.7%
Games	568	3.6%
Art & Design	555	3.5%
Travel	540	3.4%
History	534	3.4%
Fashion & Beauty	504	3.2%

Formats

Format	Rows	Share
Personal Blog	2,984	19.0%
News Article	2,520	16.1%
Nonfiction Writing	1,342	8.6%
News (Org.)	638	4.1%
About (Pers.)	482	3.1%
Knowledge Article	481	3.1%
Tutorial	481	3.1%
User Review	481	3.1%
Product Page	479	3.1%
Documentation	476	3.0%
Customer Support	475	3.0%
Creative Writing	471	3.0%
FAQ	470	3.0%
Legal Notices	470	3.0%
Academic Writing	466	3.0%
About (Org.)	464	3.0%
Listicle	463	2.9%
Comment Section	453	2.9%
Q&A Forum	443	2.8%
Audio Transcript	413	2.6%
Structured Data	391	2.5%
Content Listing	352	2.2%

Evaluation protocol

Classes

original is Human, final_response is AI. Each detector scores both texts of every row.

Thresholds

All 15,695 rows are used; there is no sweep/evaluation split. Each detector's thresholds are placed exactly on its human scores so that 1% and 0.1% of human texts are called AI. 9 detectors call lower scores AI (for example, perplexity).

Metrics

N (texts scored), AUROC, and TPR at the 1% and 0.1% FPR thresholds. Subsets are scored at the corpus-wide thresholds rather than refitted.

Filters

None. No cosine, soft-ngram, Jaccard, or other analysis-time filtering is applied.

Skipped

None

Config

config/analysis.toml

Schema & usage57 columns · loading snippet⌄

57 columns per row. Detector columns come in pairs, one per text of the row.

Text

original

Human-written source (class: Human)

final_response

Generated text (class: AI)

response_0

Generator output for the first turn

response_1

Generator output for the second turn (empty for single-turn prompts)

Generation

prompt

struct: chat_turns (list of strings), use_multiturn, examples, metadata.PROMPT_TYPE

generator_model

Hugging Face id of the generator model

generation_params

JSON string of sampling parameters

Metadata

topic

Topic of the human source

format

Format of the human source

Distances

jaccard_1, jaccard_2, levenshtein, softngram, cosdist, bertscore, bertscore_precision, bertscore_recall, moverscore, reranker

10 float columns measuring how far the rewrite moved from the original

Other columns

original_editlens_bucket_roberta_large, final_response_editlens_bucket_roberta_large, original_editlens_bucket_llama_3_2_3b, final_response_editlens_bucket_llama_3_2_3b, original_editlens_bucket_giga_roberta, final_response_editlens_bucket_giga_roberta, original_editlens_bucket_giga_llama_3_2_3b, final_response_editlens_bucket_giga_llama_3_2_3b

Not used by this report

Detector scores

One float column per text: {original|final_response}<suffix>

_editlens_score_roberta_large

EditLens Roberta-Large Score

_editlens_score_llama_3_2_3b

EditLens Llama-3.2-3B Score

_editlens_score_giga_roberta

Giga EditLens Roberta Score

_editlens_score_giga_llama_3_2_3b

Giga EditLens Llama-3.2-3B Score

_perplexity_llama_instruct

Perplexity (Llama-3.2-3B-Instruct)

_perplexity_llama_base

Perplexity (Llama-3.2-3B)

_entropy_llama_instruct

Entropy (Llama-3.2-3B-Instruct)

_entropy_llama_base

Entropy (Llama-3.2-3B)

_topp_outlier_llama_instruct

Top-p Outliers (Llama-3.2-3B-Instruct)

_topp_outlier_llama_base

Top-p Outliers (Llama-3.2-3B)

_topk_outlier_llama_instruct

Top-k Outliers (Llama-3.2-3B-Instruct)

_topk_outlier_llama_base

Top-k Outliers (Llama-3.2-3B)

_fastdetectgpt_llama_instruct

FastDetectGPT (Llama-3.2-3B-Instruct)

_fastdetectgpt_llama_base

FastDetectGPT (Llama-3.2-3B)

_binoculars

Binoculars

Load it

from datasets import concatenate_datasets, get_dataset_config_names, load_dataset

repo = "G-reen/fastdetector-test-stat-test"
# every generator config is its own dataset config
ds = concatenate_datasets([
    load_dataset(repo, config, split="train")
    for config in get_dataset_config_names(repo)
])

Univariate statistics40 statistics · all 15,695 rows valid⌄

Every statistic the report does arithmetic on, over the complete corpus. Missing or non-finite values are counted as invalid.

Statistic	Mean	Median	Std	Min	Max	Invalid
jaccard_1	0.5746	0.6719	0.2821	0.0000	1.0000	0
jaccard_2	0.6923	0.8358	0.3095	0.0180	1.0000	0
levenshtein	2123.5641	1533.0000	2534.1196	3.0000	79984.0000	0
softngram	0.5301	0.5883	0.3602	0.0000	1.0000	0
cosdist	0.1589	0.0863	0.1807	-0.0072	0.9648	0
bertscore	0.1181	0.1238	0.0733	0.0012	0.7966	0
bertscore_precision	0.1170	0.1236	0.0719	0.0012	0.7497	0
bertscore_recall	0.1184	0.1218	0.0786	0.0006	0.8287	0
moverscore	0.4858	0.5403	0.2000	0.0595	1.3348	0
reranker	-5.8904	-7.3125	4.0837	-10.9375	17.1250	0
EditLens Roberta-Large Score (Human)	0.0463	0.0204	0.0716	0.0064	0.9838	0
EditLens Roberta-Large Score (AI)	0.3628	0.2611	0.3401	0.0064	0.9996	0
EditLens Llama-3.2-3B Score (Human)	0.0299	0.0159	0.0406	0.0003	0.4905	0
EditLens Llama-3.2-3B Score (AI)	0.3439	0.2804	0.3100	0.0004	1.0000	0
Giga EditLens Roberta Score (Human)	0.0086	0.0038	0.0258	0.0013	0.6720	0
Giga EditLens Roberta Score (AI)	0.3834	0.2360	0.3997	0.0013	1.0000	0
Giga EditLens Llama-3.2-3B Score (Human)	0.0050	0.0026	0.0114	0.0005	0.1835	0
Giga EditLens Llama-3.2-3B Score (AI)	0.3285	0.1771	0.3631	0.0006	1.0000	0
Perplexity (Llama-3.2-3B-Instruct) (Human)	15.5733	13.9608	6.9930	2.9954	65.3062	0
Perplexity (Llama-3.2-3B-Instruct) (AI)	15.5244	13.6347	11.0491	1.0071	537.9031	0
Perplexity (Llama-3.2-3B) (Human)	11.8289	10.8280	4.9417	1.2068	47.0632	0
Perplexity (Llama-3.2-3B) (AI)	12.4909	11.0577	8.9863	1.0034	612.3403	0
Entropy (Llama-3.2-3B-Instruct) (Human)	2.4983	2.4877	0.4347	1.0731	4.2297	0
Entropy (Llama-3.2-3B-Instruct) (AI)	2.4463	2.4341	0.4871	0.0365	7.1878	0
Entropy (Llama-3.2-3B) (Human)	2.3785	2.3792	0.3802	0.2294	3.7963	0
Entropy (Llama-3.2-3B) (AI)	2.4149	2.4143	0.3978	0.0100	5.5116	0
Top-p Outliers (Llama-3.2-3B-Instruct) (Human)	0.0550	0.0537	0.0120	0.0215	0.1231	0
Top-p Outliers (Llama-3.2-3B-Instruct) (AI)	0.0552	0.0537	0.0142	0.0000	0.2083	0
Top-p Outliers (Llama-3.2-3B) (Human)	0.0421	0.0417	0.0094	0.0064	0.0796	0
Top-p Outliers (Llama-3.2-3B) (AI)	0.0398	0.0393	0.0117	0.0000	0.1818	0
Top-k Outliers (Llama-3.2-3B-Instruct) (Human)	0.0960	0.0917	0.0346	0.0064	0.2630	0
Top-k Outliers (Llama-3.2-3B-Instruct) (AI)	0.0957	0.0912	0.0406	0.0001	0.5000	0
Top-k Outliers (Llama-3.2-3B) (Human)	0.0778	0.0739	0.0306	0.0039	0.2192	0
Top-k Outliers (Llama-3.2-3B) (AI)	0.0814	0.0764	0.0372	0.0001	0.5000	0
FastDetectGPT (Llama-3.2-3B-Instruct) (Human)	-1.8366	-1.7972	1.6611	-10.5708	5.9897	0
FastDetectGPT (Llama-3.2-3B-Instruct) (AI)	-1.9261	-1.9196	1.8921	-14.4647	56.7824	0
FastDetectGPT (Llama-3.2-3B) (Human)	-0.1385	-0.0941	1.0285	-5.8795	2.7665	0
FastDetectGPT (Llama-3.2-3B) (AI)	0.1040	0.0356	1.5462	-10.4853	16.7697	0
Binoculars (Human)	0.8469	0.8544	0.0681	0.0379	1.0063	0
Binoculars (AI)	0.8690	0.8694	0.0622	0.0355	1.2666	0

Correlation heatmappairwise Pearson r⌄

Each pair is correlated over the rows where both are finite.Open full size ↗

Generated by the FastDetector analysis pipeline from the complete published corpus.config/analysis.toml