Datasets:
Modalities:
Text
Formats:
text
Languages:
English
Size:
< 1K
Tags:
ai-text-detection
forensic-fingerprinting
stylometrics
cloze-congruence
model-provenance
ed25519-seal
License:
Upload benchmark evaluation report: pratyushpuri_2025_evaluation_report.txt
Browse files
dataset_evaluations/pratyushpuri_2025_evaluation_report.txt
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
===============================================================================================
|
| 2 |
+
BENCHMARK EVALUATION: pratyushpuri/ai-vs-human-content-detection-1000-record-in-2025
|
| 3 |
+
Source: https://www.kaggle.com/datasets/pratyushpuri/ai-vs-human-content-detection-1000-record-in-2025
|
| 4 |
+
Records: 1,367 rows (684 Human [0] vs 683 AI [1]) across 8 content types
|
| 5 |
+
===============================================================================================
|
| 6 |
+
2026-09-10 22:19:32.121 WARNING streamlit.runtime.scriptrunner_utils.script_run_context: Thread 'MainThread': missing ScriptRunContext! This warning can be ignored when running in bare mode.
|
| 7 |
+
2026-09-10 22:19:32.121 WARNING streamlit.runtime.state.session_state_proxy: Session state does not function when running a script without `streamlit run`
|
| 8 |
+
2026-09-10 22:19:32.121 WARNING streamlit.runtime.scriptrunner_utils.script_run_context: Thread 'MainThread': missing ScriptRunContext! This warning can be ignored when running in bare mode.
|
| 9 |
+
-----------------------------------------------------------------------------------------------
|
| 10 |
+
[STAGE 1/2] Evaluating 10 Human-Labeled Records (False Accusation Rate Test):
|
| 11 |
+
-----------------------------------------------------------------------------------------------
|
| 12 |
+
Human #01 | Type=essay | Words=107 | P(AI)= 2.0% | PASS [ZERO FAR]
|
| 13 |
+
Human #02 | Type=product_review | Words= 48 | P(AI)= 2.0% | PASS [ZERO FAR]
|
| 14 |
+
Human #03 | Type=creative_writing | Words= 79 | P(AI)= 2.0% | PASS [ZERO FAR]
|
| 15 |
+
Human #04 | Type=news_article | Words=202 | P(AI)= 2.0% | PASS [ZERO FAR]
|
| 16 |
+
Human #05 | Type=social_media | Words= 7 | P(AI)= 2.0% | PASS [ZERO FAR]
|
| 17 |
+
Human #06 | Type=article | Words= 40 | P(AI)= 2.0% | PASS [ZERO FAR]
|
| 18 |
+
Human #07 | Type=news_article | Words= 77 | P(AI)= 2.0% | PASS [ZERO FAR]
|
| 19 |
+
Human #08 | Type=news_article | Words=158 | P(AI)= 2.0% | PASS [ZERO FAR]
|
| 20 |
+
Human #09 | Type=article | Words=170 | P(AI)= 2.0% | PASS [ZERO FAR]
|
| 21 |
+
Human #10 | Type=essay | Words= 86 | P(AI)= 2.0% | PASS [ZERO FAR]
|
| 22 |
+
|
| 23 |
+
>> Human False Accusation Rate: 0.00% (0/10 false positives)
|
| 24 |
+
|
| 25 |
+
-----------------------------------------------------------------------------------------------
|
| 26 |
+
[STAGE 2/2] Evaluating 10 AI-Labeled Records:
|
| 27 |
+
-----------------------------------------------------------------------------------------------
|
| 28 |
+
AI #01 | Type=academic_paper | Words=288 | P(AI)= 2.0% | Attr=Anthropic Claude 3.5 Sonnet | REVIEW
|
| 29 |
+
AI #02 | Type=essay | Words=253 | P(AI)= 2.0% | Attr=Anthropic Claude 3.5 Sonnet | REVIEW
|
| 30 |
+
AI #03 | Type=academic_paper | Words=420 | P(AI)= 2.0% | Attr=Anthropic Claude 3.5 Sonnet | REVIEW
|
| 31 |
+
AI #04 | Type=creative_writing | Words=196 | P(AI)= 2.0% | Attr=Anthropic Claude 3.5 Sonnet | REVIEW
|
| 32 |
+
AI #05 | Type=news_article | Words=160 | P(AI)= 2.0% | Attr=Anthropic Claude 3.5 Sonnet | REVIEW
|
| 33 |
+
AI #06 | Type=blog_post | Words=198 | P(AI)= 2.0% | Attr=Anthropic Claude 3.5 Sonnet | REVIEW
|
| 34 |
+
AI #07 | Type=blog_post | Words= 84 | P(AI)= 2.0% | Attr=Anthropic Claude 3.5 Sonnet | REVIEW
|
| 35 |
+
AI #08 | Type=article | Words= 89 | P(AI)= 2.0% | Attr=Anthropic Claude 3.5 Sonnet | REVIEW
|
| 36 |
+
AI #09 | Type=article | Words=156 | P(AI)= 2.0% | Attr=Anthropic Claude 3.5 Sonnet | REVIEW
|
| 37 |
+
AI #10 | Type=essay | Words=147 | P(AI)= 2.0% | Attr=Anthropic Claude 3.5 Sonnet | REVIEW
|
| 38 |
+
|
| 39 |
+
===============================================================================================
|
| 40 |
+
BENCHMARK PERFORMANCE SUMMARY ON pratyushpuri DATASET
|
| 41 |
+
===============================================================================================
|
| 42 |
+
Human False Accusation Rate: 0.00% (Target: 0.00%)
|
| 43 |
+
AI Detection Sensitivity: 0.00% (0/10) | Mean P(AI): 2.0%
|
| 44 |
+
===============================================================================================
|