Datasetpublic

G-reen/cc-re-2020-stat-train

Auto-Generated FastDetector Dataset Dataset: G-reen/cc-re-2020-stat-train Globals Config: config/globals_re2020.toml Analysis Config: config/analysis_exclude_deepseek.toml Rows: 108,377 Evaluation Results Prompt Subsets: 4 (direct_reference, indirect_reference, revise, rewrite) Generator Configs: 11 (Laguna-S-2.1-NVFP4 (Temp: 1.0), Llama-3.3-70B-Instruct-NVFP4 (Temp: 0.6), Llama-3.3-70B-Instruct-NVFP4 (Temp: 1.25), Mistral-Small-4-119B-2603-NVFP4 (Temp: 0.7)… See the full description on the dataset page: https://huggingface.co/datasets/G-reen/cc-re-2020-stat-train.

sourceHugging Faceupdated 14d agoView on Hugging Face

0likes550downloads

Dataset Card

Auto-Generated FastDetector Dataset

—Dataset: G-reen/cc-re-2020-stat-train
—Globals Config: config/globals_re2020.toml
—Analysis Config: config/analysis_exclude_deepseek.toml
—Rows: 108,377

Evaluation Results

—Prompt Subsets: 4 (directreference, indirectreference, revise, rewrite)
—Generator Configs: 11 (Laguna-S-2.1-NVFP4 (Temp: 1.0), Llama-3.3-70B-Instruct-NVFP4 (Temp: 0.6), Llama-3.3-70B-Instruct-NVFP4 (Temp: 1.25), Mistral-Small-4-119B-2603-NVFP4 (Temp: 0.7), Mistral-Small-4-119B-2603-NVFP4 (Temp: 1.25), Ornith-1.5-35B-A3B-NVFP4 (Temp: 0.6), Qwen3.8-27B-AWQ-INT4 (Temp: 0.7), Qwen3.8-27B-AWQ-INT4 (Temp: 1.25), gemma-4-31B-it-AWQ-4bit (Temp: 1.0), gemma-4-31B-it-AWQ-4bit (Temp: 1.25), granite-4.2-30b-nvfp4 (Temp: 1.0))
—Classifiers: 13 (EditLens Roberta-Large Score, EditLens Roberta-Large Bucket, Perplexity (Llama-3.2-3B-Instruct) [skipped], Perplexity (Llama-3.2-3B) [skipped], Entropy (Llama-3.2-3B-Instruct) [skipped], Entropy (Llama-3.2-3B) [skipped], Top-p Outliers (Llama-3.2-3B-Instruct) [skipped], Top-p Outliers (Llama-3.2-3B) [skipped], Top-k Outliers (Llama-3.2-3B-Instruct) [skipped], Top-k Outliers (Llama-3.2-3B) [skipped], FastDetectGPT (Llama-3.2-3B-Instruct) [skipped], FastDetectGPT (Llama-3.2-3B) [skipped], Binoculars [skipped])
—Filter Conditions: generator_model != deepseek-ai/DeepSeek-V4-Flash-0731
—Evaluation / Validation Rows: 97,539 / 10,838 (validation_size = 0.1)
—Base Columns: original (Human), final_response (AI)

The best classifier was EditLens Roberta-Large Score with an AUROC of 0.9044. The hardest prompt subset was rewrite with a TPR of 0.2405, and the hardest generator config was Ornith-1.5-35B-A3B-NVFP4 (Temp: 0.6) with a TPR of 0.4525.

Classifier	Threshold	AUROC	TPR	FPR	Accuracy	F1
✔️ EditLens Roberta-Large Score	0.5918	0.9044	0.4830	0.0050	0.7390	0.6492
❗ EditLens Roberta-Large Bucket	0.0000	0.8462	0.7272	0.0583	0.8345	0.8146

Classifier metrics averaged within each prompt and generator subset:

Subset	Average AUROC	Average TPR	Average FPR	Average Accuracy	Average F1
✔️ Prompt: revise	0.9670	0.7786	0.0318	0.8734	0.8516
Model: Mistral-Small-4-119B-2603-NVFP4 (Temp: 1.25)	0.9499	0.7463	0.0333	0.8565	0.8294
Model: Mistral-Small-4-119B-2603-NVFP4 (Temp: 0.7)	0.9294	0.7140	0.0344	0.8398	0.8088
Prompt: direct_reference	0.9250	0.7423	0.0314	0.8554	0.8339
Model: Laguna-S-2.1-NVFP4 (Temp: 1.0)	0.9103	0.6999	0.0321	0.8339	0.8033
Prompt: indirect_reference	0.9026	0.6460	0.0321	0.8069	0.7633
Model: granite-4.2-30b-nvfp4 (Temp: 1.0)	0.8882	0.6358	0.0324	0.8017	0.7541
Model: Qwen3.8-27B-AWQ-INT4 (Temp: 1.25)	0.8786	0.6127	0.0310	0.7908	0.7383
Model: Qwen3.8-27B-AWQ-INT4 (Temp: 0.7)	0.8660	0.6029	0.0317	0.7856	0.7310
Model: gemma-4-31B-it-AWQ-4bit (Temp: 1.25)	0.8639	0.5843	0.0303	0.7770	0.7149
Model: gemma-4-31B-it-AWQ-4bit (Temp: 1.0)	0.8621	0.5857	0.0315	0.7771	0.7159
Model: Llama-3.3-70B-Instruct-NVFP4 (Temp: 1.25)	0.8342	0.5138	0.0290	0.7424	0.6604
Model: Llama-3.3-70B-Instruct-NVFP4 (Temp: 0.6)	0.8279	0.5079	0.0325	0.7377	0.6548
Model: Ornith-1.5-35B-A3B-NVFP4 (Temp: 0.6)	0.8179	0.4525	0.0301	0.7112	0.5974
❗ Prompt: rewrite	0.7013	0.2405	0.0314	0.6045	0.3600

✔️ marks the best AUROC, ❗ the worst.

Statistics of Interest

[image] [image] [image] [image]

Appendix

Table of contents 1. Univariate Analysis 2. Correlation Heatmap 3. Distance Histograms 4. Distance Histograms per Prompt Subset 5. Distance Histograms per Generator Config Subset 6. Classifier: EditLens Roberta-Large Score - Performance: - Thresholding: - Classification Histograms: 7. Classifier: EditLens Roberta-Large Bucket - Performance: - Thresholding: - Classification Histograms:

Univariate Analysis

Every statistic the report does arithmetic on, over the 97,539-row evaluation split. Invalid counts rows whose value is missing or non-finite; those rows are excluded from the other columns.

Statistic	N	Mean	Median	Std	Min	Max
jaccard_1	97,539	0.6384	0.7649	0.2801	0.0000	1.0000
jaccard_2	97,539	0.7508	0.9132	0.3039	0.0000	1.0000
levenshtein	97,539	2344.5066	1513.0000	3557.4289	1.0000	178023.0000
cosdist	97,539	0.2051	0.1435	0.2026	-0.0072	1.0397
EditLens Roberta-Large Score (Human)	97,539	0.0580	0.0207	0.0960	0.0062	0.9996
EditLens Roberta-Large Score (AI)	97,539	0.5671	0.5574	0.3769	0.0065	0.9996
EditLens Roberta-Large Bucket (Human)	97,539	0.0686	0.0000	0.3074	0.0000	3.0000
EditLens Roberta-Large Bucket (AI)	97,539	1.6549	1.0000	1.2999	0.0000	3.0000

Correlation Heatmap

Pearson correlation between every statistic of interest, computed over the rows where both statistics are present.

[image]

Distance Histograms

[image] [image] [image] [image]

Distance Histograms per Prompt Subset

[image] [image] [image] [image]

Distance Histograms per Generator Config Subset

[image] [image] [image] [image]

Classifier: EditLens Roberta-Large Score

Performance:

Subset	N	AUROC	TPR	FPR	Accuracy	F1
Overall	195,078	0.9044	0.4830	0.0050	0.7390	0.6492
Prompt: direct_reference	49,400	0.9463	0.6521	0.0055	0.8233	0.7868
Prompt: indirect_reference	48,752	0.9390	0.5269	0.0055	0.7607	0.6877
✔️ Prompt: revise	49,374	0.9814	0.6240	0.0045	0.8098	0.7663
❗ Prompt: rewrite	47,552	0.7469	0.1160	0.0048	0.5556	0.2069
Model: Laguna-S-2.1-NVFP4 (Temp: 1.0)	17,608	0.9307	0.5911	0.0044	0.7933	0.7409
Model: Llama-3.3-70B-Instruct-NVFP4 (Temp: 0.6)	17,982	0.8779	0.4224	0.0057	0.7084	0.5916
Model: Llama-3.3-70B-Instruct-NVFP4 (Temp: 1.25)	17,834	0.8809	0.4186	0.0036	0.7075	0.5887
Model: Mistral-Small-4-119B-2603-NVFP4 (Temp: 0.7)	17,754	0.9481	0.5737	0.0050	0.7844	0.7268
Model: Mistral-Small-4-119B-2603-NVFP4 (Temp: 1.25)	17,774	0.9646	0.5900	0.0053	0.7923	0.7396
Model: Ornith-1.5-35B-A3B-NVFP4 (Temp: 0.6)	17,414	0.8619	0.3163	0.0045	0.6559	0.4790
Model: Qwen3.8-27B-AWQ-INT4 (Temp: 0.7)	17,624	0.8908	0.4888	0.0060	0.7414	0.6540
Model: Qwen3.8-27B-AWQ-INT4 (Temp: 1.25)	17,814	0.9073	0.4921	0.0058	0.7431	0.6570
Model: gemma-4-31B-it-AWQ-4bit (Temp: 1.0)	17,726	0.8864	0.4593	0.0042	0.7276	0.6277
Model: gemma-4-31B-it-AWQ-4bit (Temp: 1.25)	17,820	0.8884	0.4545	0.0052	0.7247	0.6228
Model: granite-4.2-30b-nvfp4 (Temp: 1.0)	17,728	0.9106	0.5047	0.0058	0.7495	0.6683

Thresholding:

—Direction: higher_is_ai
—Swept for fpr_0_5pct with a found threshold of 0.5918.

[image]

Classification Histograms:

[image]

Per Prompt Subset

[image] [image] [image] [image]

Per Generator Config Subset

[image] [image] [image] [image] [image] [image] [image] [image] [image] [image] [image]

Classifier: EditLens Roberta-Large Bucket

Performance:

Subset	N	AUROC	TPR	FPR	Accuracy	F1
Overall	195,078	0.8462	0.7272	0.0583	0.8345	0.8146
Prompt: direct_reference	49,400	0.9036	0.8326	0.0574	0.8876	0.8810
Prompt: indirect_reference	48,752	0.8662	0.7651	0.0587	0.8532	0.8390
✔️ Prompt: revise	49,374	0.9526	0.9333	0.0591	0.9371	0.9369
❗ Prompt: rewrite	47,552	0.6557	0.3650	0.0581	0.6535	0.5130
Model: Laguna-S-2.1-NVFP4 (Temp: 1.0)	17,608	0.8898	0.8087	0.0599	0.8744	0.8656
Model: Llama-3.3-70B-Instruct-NVFP4 (Temp: 0.6)	17,982	0.7779	0.5934	0.0594	0.7670	0.7180
Model: Llama-3.3-70B-Instruct-NVFP4 (Temp: 1.25)	17,834	0.7875	0.6089	0.0545	0.7772	0.7322
Model: Mistral-Small-4-119B-2603-NVFP4 (Temp: 0.7)	17,754	0.9106	0.8542	0.0638	0.8952	0.8908
Model: Mistral-Small-4-119B-2603-NVFP4 (Temp: 1.25)	17,774	0.9352	0.9026	0.0613	0.9206	0.9192
Model: Ornith-1.5-35B-A3B-NVFP4 (Temp: 0.6)	17,414	0.7739	0.5886	0.0558	0.7664	0.7159
Model: Qwen3.8-27B-AWQ-INT4 (Temp: 0.7)	17,624	0.8413	0.7170	0.0574	0.8298	0.8081
Model: Qwen3.8-27B-AWQ-INT4 (Temp: 1.25)	17,814	0.8499	0.7332	0.0561	0.8386	0.8196
Model: gemma-4-31B-it-AWQ-4bit (Temp: 1.0)	17,726	0.8378	0.7121	0.0589	0.8266	0.8042
Model: gemma-4-31B-it-AWQ-4bit (Temp: 1.25)	17,820	0.8394	0.7140	0.0554	0.8293	0.8071
Model: granite-4.2-30b-nvfp4 (Temp: 1.0)	17,728	0.8657	0.7668	0.0590	0.8539	0.8400

Thresholding:

—Direction: higher_is_ai
—Swept for f1 with a found threshold of 0.0000.

[image]

Classification Histograms:

[image]

Per Prompt Subset

[image] [image] [image] [image]

Per Generator Config Subset

[image] [image] [image] [image] [image] [image] [image] [image] [image] [image] [image]