original string | prompt dict | topic string | format string | response_0 string | response_1 string | generator_model string | generation_params string | final_response string | jaccard_1 float64 | jaccard_2 float64 | levenshtein float64 | softngram float64 | cosdist float64 | bertscore_precision float64 | bertscore_recall float64 | bertscore float64 | moverscore float64 | reranker float64 | original_perplexity_llama_instruct float64 | original_entropy_llama_instruct float64 | original_topp_outlier_llama_instruct float64 | original_topk_outlier_llama_instruct float64 | original_fastdetectgpt_llama_instruct float64 | original_perplexity_llama_base float64 | original_entropy_llama_base float64 | original_topp_outlier_llama_base float64 | original_topk_outlier_llama_base float64 | original_fastdetectgpt_llama_base float64 | final_response_perplexity_llama_instruct float64 | final_response_entropy_llama_instruct float64 | final_response_topp_outlier_llama_instruct float64 | final_response_topk_outlier_llama_instruct float64 | final_response_fastdetectgpt_llama_instruct float64 | final_response_perplexity_llama_base float64 | final_response_entropy_llama_base float64 | final_response_topp_outlier_llama_base float64 | final_response_topk_outlier_llama_base float64 | final_response_fastdetectgpt_llama_base float64 | original_binoculars float64 | final_response_binoculars float64 | original_editlens_bucket_roberta_large int64 | original_editlens_score_roberta_large float64 | final_response_editlens_bucket_roberta_large int64 | final_response_editlens_score_roberta_large float64 | original_editlens_bucket_llama_3_2_3b int64 | original_editlens_score_llama_3_2_3b float64 | final_response_editlens_bucket_llama_3_2_3b int64 | final_response_editlens_score_llama_3_2_3b float64 | original_editlens_bucket_giga_roberta int64 | original_editlens_score_giga_roberta float64 | final_response_editlens_bucket_giga_roberta int64 | final_response_editlens_score_giga_roberta float64 | original_editlens_bucket_giga_llama_3_2_3b int64 | original_editlens_score_giga_llama_3_2_3b float64 | final_response_editlens_bucket_giga_llama_3_2_3b int64 | final_response_editlens_score_giga_llama_3_2_3b float64 | original_detector_score_qwen35_35b_a3b float64 | final_response_detector_score_qwen35_35b_a3b float64 | original_detector_score_llama_3b float64 | final_response_detector_score_llama_3b float64 | original_detector_score_gemma3_27b float64 | final_response_detector_score_gemma3_27b float64 | original_detector_score_qwen35_35b_a3b_soft float64 | final_response_detector_score_qwen35_35b_a3b_soft float64 | original_detector_score_llama_3b_soft float64 | final_response_detector_score_llama_3b_soft float64 | original_detector_score_llama_3b_reg float64 | final_response_detector_score_llama_3b_reg float64 | original_llmjudge_score_opus55_high float64 | original_llmjudge_valid_opus55_high int64 | final_response_llmjudge_score_opus55_high float64 | final_response_llmjudge_valid_opus55_high int64 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
The TA office is a place for TAs to work quietly on assignment development, grading, and quiet TA-related work in small groups. It is also a room where TAs can go to get their own work done during the day and in the evenings. This room is a really great thing. Be sure to take advantage of it!
On a very limited basis, t... | {
"chat_turns": [
"<document>\n{{DOC}}\n</document>\n\nEdit this text by removing one section from it (the least necessary), you may only make minimal modifications to adjacent sections for better flow.\nOutput the full new text with no extra statements or commentations.\nBegin directly with the text itself. Do n... | Education & Jobs | Customer Support | The TA office is a place for TAs to work quietly on assignment development, grading, and quiet TA-related work in small groups. It is also a room where TAs can go to get their own work done during the day and in the evenings.
On a very limited basis, the room can also be reserved for grading meetings when private lab ... | gpt-5.6-sol | {"reasoning_effort": "none"} | The TA office is a place for TAs to work quietly on assignment development, grading, and quiet TA-related work in small groups. It is also a room where TAs can go to get their own work done during the day and in the evenings.
On a very limited basis, the room can also be reserved for grading meetings when private lab s... | 0.228758 | 0.334601 | 218 | 0.021978 | 0.002727 | 0.018102 | 0.040169 | 0.029261 | 0.253543 | -9 | 17.763068 | 2.680078 | 0.063333 | 0.106667 | -1.521432 | 15.051693 | 2.685623 | 0.046667 | 0.096667 | -0.202275 | 17.372215 | 2.624427 | 0.054054 | 0.104247 | -1.641493 | 14.650797 | 2.657985 | 0.050193 | 0.088803 | -0.191339 | 0.916149 | 0.928042 | 0 | 0.08452 | 0 | 0.080089 | 0 | 0.018928 | 0 | 0.02221 | 0 | 0.04103 | 1 | 0.322197 | 0 | 0.003831 | 0 | 0.004289 | 0.096025 | 0.9659 | 0.046725 | 0.396769 | 0.045016 | 0.597843 | 0.023331 | 0.534127 | 0.044019 | 0.246533 | 0.016357 | 0.05835 | 0 | 1 | 0.2 | 1 | |
Policy Against Sex Discrimination and Sexual Harassment*
Worcester State University prohibits discrimination on the basis of gender. Title IX of the Education Amendments of 1972 mandates that no person on the basis of sex may be denied benefits or otherwise discriminated against in any program or activity of the Univer... | {
"chat_turns": [
"<document>\n{{DOC}}\n</document>\n\nChange the tone of this text to be more objective and unbiased.\nOutput the full new text with no extra statements or commentations.\nBegin directly with the text itself. Do not add a title, a heading, or a label naming what you have written.",
"{{RESP_0}... | Education & Jobs | Legal Notices | Worcester State University prohibits discrimination based on sex or gender. Title IX of the Education Amendments of 1972 provides that no person may, on the basis of sex, be excluded from participation in, denied the benefits of, or subjected to discrimination under any University program or activity. These programs an... | Worcester State University prohibits discrimination based on sex or gender. Under Title IX of the Education Amendments of 1972, no person may, on the basis of sex, be excluded from participation in, denied the benefits of, or subjected to discrimination in any University program or activity. These programs and activiti... | gpt-5.6-sol | {"reasoning_effort": "none"} | Worcester State University prohibits discrimination based on sex or gender. Under Title IX of the Education Amendments of 1972, no person may, on the basis of sex, be excluded from participation in, denied the benefits of, or subjected to discrimination in any University program or activity. These programs and activiti... | 0.573171 | 0.77221 | 885 | 0.26259 | 0.049232 | 0.080426 | 0.100656 | 0.090654 | 0.446064 | -8.5 | 7.178752 | 1.597816 | 0.065789 | 0.052632 | -4.032243 | 5.582224 | 1.490615 | 0.047368 | 0.052632 | -2.659513 | 4.38475 | 1.319129 | 0.056497 | 0.033898 | -1.818851 | 3.573893 | 1.323362 | 0.028249 | 0.028249 | 0.586849 | 0.90874 | 0.792775 | 0 | 0.015375 | 0 | 0.039785 | 0 | 0.007824 | 0 | 0.082349 | 0 | 0.004752 | 0 | 0.119504 | 0 | 0.003324 | 0 | 0.072372 | 0.024238 | 0.991015 | 0.016153 | 0.967899 | 0.018978 | 0.905322 | 0.009413 | 0.560734 | 0.012432 | 0.577495 | 0.017334 | 0.217773 | 0 | 1 | 0.8 | 1 |
An Ohio judge has deemed Cleveland's school-choice program constitutional, opening the door for 2,000 low-income children to enroll in private and religious schools at state expense.
In a case being watched nationwide, Judge Lisa Sadler of Franklin County Common Pleas Court in Columbus ruled that the Cleveland voucher ... | {
"chat_turns": [
"<document>\n{{DOC}}\n</document>\n\nWrite 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.\nOutput the full new text with no extra statements or commentations.\nB... | Education & Jobs | News Article | An Ohio trial judge’s decision to uphold Cleveland’s school-voucher program may allow 2,000 low-income children to attend private and religious schools at public expense, but it does not settle the program’s constitutional or educational legitimacy. The ruling rests on the disputed premise that public aid reaching sect... | gpt-5.6-sol | {"reasoning_effort": "none"} | An Ohio trial judge’s decision to uphold Cleveland’s school-voucher program may allow 2,000 low-income children to attend private and religious schools at public expense, but it does not settle the program’s constitutional or educational legitimacy. The ruling rests on the disputed premise that public aid reaching sect... | 0.82346 | 0.947242 | 6,118 | 0.888636 | 0.118594 | 0.18737 | 0.161711 | 0.17474 | 0.651442 | -7.6875 | 9.7065 | 2.00258 | 0.056471 | 0.063529 | -4.012297 | 7.411657 | 2.05308 | 0.04 | 0.050588 | 0.778104 | 16.599679 | 2.5154 | 0.061746 | 0.10362 | -5.155605 | 12.156069 | 2.53257 | 0.042583 | 0.074521 | 0.634158 | 0.883329 | 0.876824 | 0 | 0.069764 | 0 | 0.235902 | 0 | 0.003023 | 2 | 0.482623 | 0 | 0.003143 | 1 | 0.323891 | 0 | 0.001496 | 1 | 0.44502 | 0.751283 | 0.999968 | 0.103749 | 0.999987 | 0.115561 | 0.999999 | 0.156105 | 0.685949 | 0.041462 | 0.651355 | 0.00708 | 0.386719 | 0 | 1 | 0.8 | 1 | |
Students in the Crookston, MN area, have a few schools to choose from if they want to stay within an hour's drive of the town. Based on statistics from the National Center for Educational Statistics (NCES), you can find out more details about these schools.
Closest Schools to Crookston, MN
The closest schools to Crooks... | {
"chat_turns": [
"<document>\n{{DOC}}\n</document>\n\nWrite a different piece of text on the same topic.\nOutput the full new text with no extra statements or commentations.\nBegin directly with the text itself. Do not add a title, a heading, or a label naming what you have written.\nThe final text must have the... | Education & Jobs | Structured Data | {
"location": {
"city": "Crookston",
"state": "Minnesota",
"search_radius": "Approximately one hour by car"
},
"data_source": "National Center for Education Statistics (NCES)",
"schools": [
{
"name": "University of Minnesota-Crookston",
"city": "Crookston",
"distance_miles": 0,... | gpt-5.6-sol | {"reasoning_effort": "none"} | {
"location": {
"city": "Crookston",
"state": "Minnesota",
"search_radius": "Approximately one hour by car"
},
"data_source": "National Center for Education Statistics (NCES)",
"schools": [
{
"name": "University of Minnesota-Crookston",
"city": "Crookston",
"distance_miles": 0,
"institution_type": "Public 4-year univer... | 0.906475 | 0.970787 | 1,801 | 0.737646 | 0.121328 | 0.26311 | 0.207503 | 0.236317 | 0.813583 | -8.5 | 4.698159 | 1.47541 | 0.03525 | 0.040816 | -1.012815 | 4.088899 | 1.432545 | 0.033395 | 0.03525 | 0.358044 | 3.403241 | 1.06581 | 0.027907 | 0.048837 | -2.235611 | 3.076037 | 1.104553 | 0.027907 | 0.034884 | -0.268207 | 0.782973 | 0.878122 | 0 | 0.041768 | 0 | 0.230631 | 0 | 0.088305 | 3 | 0.663443 | 0 | 0.007384 | 3 | 0.990639 | 0 | 0.018356 | 3 | 0.66342 | 0.127653 | 0.999964 | 0.053403 | 0.989986 | 0.046378 | 0.999974 | 0.033086 | 0.70253 | 0.046725 | 0.577019 | 0.011169 | 0.472656 | 0.2 | 1 | 0.8 | 1 | |
Deezer has announced a partnership with Feature.fm, the advertising network through which artists can promote their music inside streaming services.
Available to all artists, the scheme enables them to advertise their songs as sponsored content within Deezer’s free-to-use platform, guaranteeing that a targeted audience... | {
"chat_turns": [
"<document>\n{{DOC}}\n</document>\n\nWrite 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.\nOutput the full new text with no extra statements or commentations.\nBegin directly with the text itself. Do not... | Entertainment | News Article | Deezer has ended its partnership with Feature.fm, the advertising network through which artists promoted their music inside streaming services.
Effective immediately, the scheme will no longer enable artists to advertise songs as sponsored content within Deezer’s free-to-use platform. The streaming company said it wou... | gpt-5.6-sol | {"reasoning_effort": "none"} | Deezer has ended its partnership with Feature.fm, the advertising network through which artists promoted their music inside streaming services.
Effective immediately, the scheme will no longer enable artists to advertise songs as sponsored content within Deezer’s free-to-use platform. The streaming company said it woul... | 0.765625 | 0.906687 | 1,621 | 0.800246 | 0.121249 | 0.163641 | 0.142361 | 0.153135 | 0.589434 | -8.75 | 14.771809 | 2.61606 | 0.049107 | 0.095982 | -0.737565 | 11.780285 | 2.489845 | 0.049107 | 0.087054 | 0.24095 | 18.402299 | 2.599656 | 0.051044 | 0.12993 | -3.01142 | 16.242179 | 2.651046 | 0.039443 | 0.113689 | -1.348592 | 0.850034 | 0.951917 | 0 | 0.150473 | 3 | 0.99897 | 0 | 0.058515 | 1 | 0.45857 | 0 | 0.002166 | 3 | 0.999963 | 0 | 0.00548 | 1 | 0.314212 | 0.244363 | 0.999969 | 0.085709 | 0.999997 | 0.057071 | 0.999996 | 0.064653 | 0.747616 | 0.048858 | 0.639736 | 0.028076 | 0.402344 | 0 | 1 | 0.5 | 0 | |
Oprah is quitting. She's been on the air for 25 years, and it literally never once occurred to me that at some point she would probably get tired of being on the air every freaking day. The woman has 70 bajillion dollars and awesome dogs and a best friend who gets to go on the red carpet for reasons I don't understand... | {
"chat_turns": [
"<document>\n{{DOC}}\n</document>\n\nCreate a prompt that might cause an LLM to generate an output resembling this text.\nDo not output anything besides what you were requested to write, and do not output any extra commentary.",
"{{RESP_0}}\nOutput only the generated text with no extra comme... | Entertainment | Personal Blog | Write a first-person comedic blog post reacting to Oprah Winfrey’s retirement after 25 years on television. Use an irreverent, affectionate, conversational voice with hyperbole, rhetorical questions, mild profanity, celebrity references, and playful admiration of her wealth, influence, friendships, magazine, and abilit... | Oprah is retiring after 25 years on television, which means daytime TV is about to lose its sun, moon, weather system, and primary distributor of free cars.
Twenty-five years! That’s not a talk-show run; that’s a dynasty with throw pillows. Oprah interviewed everybody, launched half the people currently giving life ad... | gpt-5.6-sol | {"reasoning_effort": "none"} | Oprah is retiring after 25 years on television, which means daytime TV is about to lose its sun, moon, weather system, and primary distributor of free cars.
Twenty-five years! That’s not a talk-show run; that’s a dynasty with throw pillows. Oprah interviewed everybody, launched half the people currently giving life adv... | 0.85037 | 0.97156 | 2,951 | 0.962223 | 0.169669 | 0.184808 | 0.165583 | 0.175308 | 0.64491 | -8.5 | 21.581795 | 2.967445 | 0.044293 | 0.126065 | -1.061152 | 16.810119 | 2.832765 | 0.049404 | 0.122658 | 0.11591 | 23.873664 | 3.111199 | 0.054715 | 0.151339 | -0.763351 | 21.887707 | 3.087384 | 0.032596 | 0.152503 | 0.018184 | 0.830358 | 0.875863 | 0 | 0.098918 | 1 | 0.375217 | 0 | 0.012939 | 1 | 0.588081 | 0 | 0.014728 | 1 | 0.314288 | 0 | 0.002955 | 1 | 0.354285 | 0.032101 | 0.999981 | 0.01017 | 0.99998 | 0.017177 | 1 | 0.007816 | 0.713032 | 0.016915 | 0.63341 | 0.008545 | 0.365234 | 0 | 1 | 0.8 | 1 |
Whether you’re a herbivore, omnivore or something in between, everyone should be aware of their B12 intake.
What’s the deal with B12?
B12 is one of the cornerstones of life but this essential vitamin often receives some bad press in association with vegetarian and vegan diets. One of the biggest myths is that B12 comes... | {
"chat_turns": [
"<document>\n{{DOC}}\n</document>\n\nChange the tone of this text to be more inspiring and motivational.\nOutput the full new text with no extra statements or commentations.\nBegin directly with the text itself. Do not add a title, a heading, or a label naming what you have written.\nThe final t... | Health | Knowledge Article | Whether you’re a herbivore, omnivore, or somewhere in between, understanding your vitamin B12 needs is a positive step towards protecting your long-term health.
B12 is one of the cornerstones of life. Although this essential vitamin is often discussed in relation to vegetarian and vegan diets, it is important to remem... | gpt-5.6-sol | {"reasoning_effort": "none"} | Whether you’re a herbivore, omnivore, or somewhere in between, understanding your vitamin B12 needs is a positive step towards protecting your long-term health.
B12 is one of the cornerstones of life. Although this essential vitamin is often discussed in relation to vegetarian and vegan diets, it is important to rememb... | 0.618454 | 0.80536 | 1,446 | 0.393805 | 0.076109 | 0.106453 | 0.086322 | 0.0965 | 0.443763 | -6.875 | 8.108026 | 2.059097 | 0.034722 | 0.045139 | -0.417296 | 6.38799 | 1.923687 | 0.029514 | 0.029514 | 0.923929 | 7.903368 | 1.935988 | 0.042003 | 0.059774 | -1.723123 | 6.607204 | 1.920006 | 0.040388 | 0.050081 | 0.443283 | 0.788072 | 0.847413 | 0 | 0.01105 | 0 | 0.191521 | 0 | 0.016433 | 1 | 0.294987 | 0 | 0.003844 | 3 | 0.814639 | 0 | 0.002696 | 1 | 0.414758 | 0.117572 | 0.999962 | 0.030676 | 0.999997 | 0.062789 | 1 | 0.013637 | 0.679179 | 0.046034 | 0.581778 | 0.004181 | 0.271484 | 0.2 | 1 | 0.8 | 1 | |
“A THOUSAND shall fall beside thee, and ten thousand at thy right hand: but it shall not come nigh thee.”
The promise of immunity in Psalm 91 was sung by the choir of Westminster Abbey on Thursday night last week, on the eve of the centenary of the Battle of the Somme. By the end of 1 July 1916, nearly 20,000 Allied so... | {
"chat_turns": [
"<document>\n{{DOC}}\n</document>\n\nEdit this text by removing one section from it (the least necessary), you may only make minimal modifications to adjacent sections for better flow.\nOutput the full new text with no extra statements or commentations.\nBegin directly with the text itself. Do n... | History | News Article | “A THOUSAND shall fall beside thee, and ten thousand at thy right hand: but it shall not come nigh thee.”
The promise of immunity in Psalm 91 was sung by the choir of Westminster Abbey on Thursday night last week, on the eve of the centenary of the Battle of the Somme. By the end of 1 July 1916, nearly 20,000 Allied s... | “A THOUSAND shall fall beside thee, and ten thousand at thy right hand: but it shall not come nigh thee.”
Psalm 91’s promise of protection rang through Westminster Abbey on Thursday night last week, during a service held on the eve of the centenary of the Battle of the Somme. By the end of 1 July 1916, nearly 20,000 A... | gpt-5.6-sol | {"reasoning_effort": "none"} | “A THOUSAND shall fall beside thee, and ten thousand at thy right hand: but it shall not come nigh thee.”
Psalm 91’s promise of protection rang through Westminster Abbey on Thursday night last week, during a service held on the eve of the centenary of the Battle of the Somme. By the end of 1 July 1916, nearly 20,000 Al... | 0.313356 | 0.444894 | 903 | 0.130466 | 0.026878 | 0.030471 | 0.031834 | 0.031153 | 0.230431 | -9.125 | 13.313351 | 2.627403 | 0.038817 | 0.098891 | 0.572964 | 10.096187 | 2.277007 | 0.046211 | 0.083179 | -0.564298 | 13.075548 | 2.493509 | 0.050524 | 0.102002 | -1.146074 | 9.48722 | 2.210489 | 0.045758 | 0.081983 | -0.635167 | 0.769799 | 0.779108 | 0 | 0.04378 | 0 | 0.070839 | 0 | 0.012526 | 0 | 0.065241 | 0 | 0.011258 | 0 | 0.008758 | 0 | 0.002061 | 0 | 0.005309 | 0.079782 | 0.830044 | 0.03138 | 0.389663 | 0.076408 | 0.504883 | 0.019124 | 0.250913 | 0.023331 | 0.163445 | 0.004639 | 0.02124 | 0 | 1 | 0.2 | 1 |
The army was tight-lipped, saying only that after completing the tests it had handed over a "safe parcel of land" to Walbrzych municipal authorities.
"The army's job is done as we're not in the business of treasure hunting," Poland's Defence Minister Tomasz Siemoniak said, quoted by the Polish PAP news agency.
Both mil... | {
"chat_turns": [
"<document>\n{{DOC}}\n</document>\n\nRewrite 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.\nOutput the full new text with no extra statements or commentations.\nBegin ... | History | News Article | The army remained guarded, revealing only that, once its examination was finished, the area was deemed secure and returned to the municipal authorities in Walbrzych.
"The army's job is done as we're not in the business of treasure hunting," Poland's Defence Minister Tomasz Siemoniak said, quoted by the Polish PAP news... | gpt-5.6-sol | {"reasoning_effort": "none"} | The army remained guarded, revealing only that, once its examination was finished, the area was deemed secure and returned to the municipal authorities in Walbrzych.
"The army's job is done as we're not in the business of treasure hunting," Poland's Defence Minister Tomasz Siemoniak said, quoted by the Polish PAP news ... | 0.085603 | 0.119565 | 98 | 0.049796 | 0.003856 | 0.010248 | 0.012028 | 0.011139 | 0.129211 | -8.1875 | 15.171564 | 2.396867 | 0.044681 | 0.095745 | -3.291775 | 10.677485 | 2.285475 | 0.02766 | 0.076596 | -0.888934 | 15.606604 | 2.521164 | 0.034115 | 0.100213 | -2.25361 | 10.940652 | 2.3323 | 0.027719 | 0.078891 | -0.645416 | 0.858351 | 0.825829 | 0 | 0.013228 | 0 | 0.013538 | 0 | 0.008919 | 0 | 0.012037 | 0 | 0.001723 | 0 | 0.001801 | 0 | 0.0012 | 0 | 0.001491 | 0.154058 | 0.2227 | 0.057071 | 0.087564 | 0.035411 | 0.104112 | 0.061876 | 0.090093 | 0.03259 | 0.071591 | 0.006592 | 0.013672 | 0 | 1 | 0 | 1 | |
Animal welfare is increasing in importance in agricultural systems in many countries around the world. As the 13th edition of the IDF Animal Health Report is released, Louise Winblad of Växa Sverige reflects on why information on animal welfare should be communicated globally, and shares details on some of the case stu... | {
"chat_turns": [
"<document>\n{{DOC}}\n</document>\n\nClarify this text by defining any terms that might be unclear to the reader.\nOutput the full new text with no extra statements or commentations.\nBegin directly with the text itself. Do not add a title, a heading, or a label naming what you have written.",
... | Industrial | Knowledge Article | Animal welfare—the physical and mental well-being of animals—is becoming increasingly important in agricultural systems in many countries. With the release of the 13th edition of the International Dairy Federation (IDF) Animal Health Report, Louise Winblad of Växa Sverige, a Swedish organization that provides advice an... | Animal welfare, understood as the physical and psychological well-being of animals, has assumed increasing importance within agricultural systems worldwide. Following the publication of the 13th edition of the International Dairy Federation (IDF) Animal Health Report, Louise Winblad of Växa Sverige, a Swedish organizat... | gpt-5.6-sol | {"reasoning_effort": "none"} | Animal welfare, understood as the physical and psychological well-being of animals, has assumed increasing importance within agricultural systems worldwide. Following the publication of the 13th edition of the International Dairy Federation (IDF) Animal Health Report, Louise Winblad of Växa Sverige, a Swedish organizat... | 0.798859 | 0.932927 | 5,078 | 0.706021 | 0.090545 | 0.19149 | 0.141663 | 0.167321 | 0.631698 | -8.75 | 22.97677 | 2.990995 | 0.058637 | 0.137876 | -1.585891 | 17.479132 | 2.738711 | 0.055468 | 0.115689 | -1.449263 | 10.45818 | 2.207672 | 0.050847 | 0.066102 | -2.44508 | 8.743706 | 2.276833 | 0.037288 | 0.058475 | 1.961352 | 0.880632 | 0.87831 | 0 | 0.011493 | 1 | 0.268389 | 0 | 0.003894 | 3 | 0.668651 | 0 | 0.002349 | 1 | 0.366832 | 0 | 0.000983 | 1 | 0.242613 | 0.051845 | 0.999979 | 0.018547 | 0.999995 | 0.026759 | 1 | 0.008577 | 0.642432 | 0.012821 | 0.566978 | 0.004059 | 0.222656 | 0 | 1 | 0.8 | 1 |
Our Vitrified Diamond grinding wheels are proven winners for grinding difficult materials in sensitive applications like flame sprayed materials and PCD & PCBN.
We divide our vitrified diamond product line into 2 categories: Ceramics and Diamond.
For ceramic based materials our HP bond system is the right choice. This ... | {
"chat_turns": [
"<document>\n{{DOC}}\n</document>\n\nEnvision the scenario in which the author wrote this text. Describe that scenario in exhaustive detail.\nDo not output anything besides what you were requested to write, and do not output any extra commentary.",
"{{RESP_0}}\nAbove is an AI generated descr... | Industrial | Product Page | The author is likely a technical marketing writer, sales engineer, or product specialist at a manufacturer of superabrasive grinding wheels. They are preparing product-page or brochure copy aimed at industrial customers who grind exceptionally hard, brittle, or expensive materials—particularly ceramics, flame-sprayed c... | Our vitrified diamond wheels are proven winners in some of the most demanding grinding applications, including ceramics, flame-sprayed coatings, PCD and PCBN.
For ceramic-based materials, our HP bond system offers fast stock removal and excellent grind ratios. HP-bond wheels can be rotary dressed and are available in ... | gpt-5.6-sol | {"reasoning_effort": "none"} | Our vitrified diamond wheels are proven winners in some of the most demanding grinding applications, including ceramics, flame-sprayed coatings, PCD and PCBN.
For ceramic-based materials, our HP bond system offers fast stock removal and excellent grind ratios. HP-bond wheels can be rotary dressed and are available in c... | 0.696133 | 0.896923 | 2,025 | 0.775652 | 0.059533 | 0.151049 | 0.134181 | 0.142698 | 0.58811 | -8.875 | 19.985148 | 2.892825 | 0.058824 | 0.125668 | -0.860853 | 15.334753 | 2.79505 | 0.040107 | 0.09893 | 0.57183 | 11.706931 | 2.281146 | 0.046468 | 0.074349 | -1.945346 | 9.5982 | 2.315038 | 0.035316 | 0.055762 | 0.609737 | 0.870993 | 0.896403 | 0 | 0.123065 | 0 | 0.03311 | 0 | 0.014686 | 0 | 0.06547 | 0 | 0.003965 | 0 | 0.001651 | 0 | 0.007 | 0 | 0.009811 | 0.044681 | 0.999947 | 0.032837 | 0.998633 | 0.046725 | 0.999196 | 0.012432 | 0.639736 | 0.03622 | 0.518668 | 0.011047 | 0.114258 | 0 | 1 | 0.8 | 1 |
End of preview. Expand in Data Studio
Auto-generated FastDetector dataset
Fastdetector Test Stat Test Mini
Best detector
Trained detector: Qwen3.5-35B-A3B (QLoRA-style LoRA, bf16 base)
0.7909 TPR @ 1% FPRHardest prompt subset
rewrite
0.4545 max detector TPR @ 1% FPRHardest generator config
claude-opus-5-5 (Temp: Unknown)
0.6000 max detector TPR @ 1% FPR220
rows
11
generator configs
4
prompt subsets
22
detectors
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 per-subset table, threshold sweep, its TPR as AI rows with low cosdist are dropped or relabelled human, and its score distributions.
02
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.
03
Appendix
Reference material: how the dataset was built and evaluated, its schema and loading code, univariate statistics, cross-statistic correlations, and the link to the per-model analytics page.
How the dataset was built11 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 | 20 | 9.1% |
| claude-opus-5 | — | type disabled | 20 | 9.1% |
| claude-opus-5-5 | — | effort low | 20 | 9.1% |
| gemini-3.1-pro-preview | — | thinking_level low | 20 | 9.1% |
| gemini-3.8-flash | — | thinking_level low | 20 | 9.1% |
| gpt-5.4 | — | reasoning_effort none | 20 | 9.1% |
| gpt-5.6-sol | — | reasoning_effort none | 20 | 9.1% |
| x-ai/grok-4.3 | — | enabled False · max_completion_tokens 16000 | 20 | 9.1% |
| moonshotai/kimi-k3 | — | enabled False · max_completion_tokens 16000 | 20 | 9.1% |
| minimax/minimax-m3 | — | enabled False · max_completion_tokens 16000 | 20 | 9.1% |
| qwen/qwen3.7-max | — | enabled False · max_completion_tokens 16000 | 20 | 9.1% |
| Total | 220 | 100% |
Prompt subsets
Instructions are grouped by their first message. Counts cover all 220 rows.
direct_reference55 rowssingle-turn · 38 instructions
Show the 38 instructions ▾
- 3× 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.
- 3× 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.
- 3× 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>>.
- 3× Write another piece of text that resembles the style and language of this text. The final text must have the topic <<topic>>.
- 2× Take a subpoint within this text and write a full new text about it, using the language and wording from the text as appropriate.
- 2× 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>>.
- 2× 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>>.
- 2× 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.
- 2× Write a different piece of text on the same topic. The final text must have the format <<format>>.
- 2× 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>>.
- 2× Write another piece of text that is of a similar theme or background as this text. The final text must have the format <<format>>.
- 2× Write another piece of text that resembles the style and language of this text.
- 2× 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>>.
- 1× 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>>.
- 1× 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>>.
- 1× 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>>.
- 1× 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>>.
- 1× 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>>.
- 1× 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.
- 1× 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>>.
- 1× Write a different piece of text on the same topic.
- 1× Write a different piece of text on the same topic. The final text must have the format <<format>> and topic <<topic>>.
- 1× Write a different piece of text on the same topic. The final text must have the topic <<topic>>.
- 1× 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>>.
- 1× Write a new piece of text that uses this text as a citation or reference.
- 1× Write a new piece of text that uses this text as a citation or reference. The final text must have the topic <<topic>>.
- 1× 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.
- 1× 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>>.
- 1× 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>>.
- 1× 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>>.
- 1× 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.
- 1× 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>>.
- 1× 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>>.
- 1× Write an alternate version of this text, it should follow the style and format of the original text.
- 1× 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>>.
- 1× 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>>.
- 1× 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>>.
- 1× Write another piece of text that is of a similar theme or background as this text. The final text must have the topic <<topic>>.
indirect_reference55 rowstwo-turn · 22 instructions
Show the 22 instructions ▾
- 6× 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.
- 4× 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.
- 4× 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.
- 4× 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.
- 4× 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.
- 4× 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.
- 3× 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.
- 3× Translate the given text to French. Do not output anything besides what you were requested to write, and do not output any extra commentary.
- 3× 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.
- 3× 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.
- 2× 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.
- 2× 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.
- 2× 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.
- 2× Translate the given text to German. Do not output anything besides what you were requested to write, and do not output any extra commentary.
- 2× 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.
- 1× 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.
- 1× 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.
- 1× 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.
- 1× Translate the given text to Chinese. Do not output anything besides what you were requested to write, and do not output any extra commentary.
- 1× Translate the given text to Spanish. Do not output anything besides what you were requested to write, and do not output any extra commentary.
- 1× 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.
- 1× 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.
revise55 rows32 single · 23 two-turn · 46 instructions
Show the 46 instructions ▾
- 3× Change the tone of this text to be more conversational and approachable.
- 3× Clarify this text by defining any terms that might be unclear to the reader.
- 2× Adapt this text for a 5th-grade reading level.
- 2× Change the tone of this text to be more lighthearted and humorous.
- 2× Change the tone of this text to be more professional and authoritative.
- 2× Clarify this text by making the cause-and-effect relationships more obvious.
- 2× Edit this text to trim the length without losing important information or tone.
- 1× Adapt this text for a 5th-grade reading level. The final text must have the topic <<topic>>.
- 1× Adapt this text for a lay audience with no technical background.
- 1× Adapt this text for a teenage audience. The final text must have the format <<format>> and topic <<topic>>.
- 1× 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>>.
- 1× Adapt this text for social media sharing.
- 1× Adapt this text for social media sharing. The final text must have the topic <<topic>>.
- 1× Change the tone of this text to be more analytical and logical.
- 1× Change the tone of this text to be more direct and bold.
- 1× Change the tone of this text to be more direct and bold. The final text must have the topic <<topic>>.
- 1× Change the tone of this text to be more empathetic and understanding.
- 1× Change the tone of this text to be more empathetic and understanding. The final text must have the format <<format>> and topic <<topic>>.
- 1× Change the tone of this text to be more enthusiastic and energetic.
- 1× Change the tone of this text to be more inspiring and motivational.
- 1× Change the tone of this text to be more inspiring and motivational. The final text must have the format <<format>>.
- 1× Change the tone of this text to be more objective and unbiased.
- 1× Change the tone of this text to be more persuasive and convincing. The final text must have the topic <<topic>>.
- 1× Change the tone of this text to be more polite and diplomatic. The final text must have the format <<format>> and topic <<topic>>.
- 1× Change the tone of this text to be more relaxed and friendly.
- 1× Change the tone of this text to be more relaxed and friendly. The final text must have the format <<format>> and topic <<topic>>.
- 1× Change the tone of this text to be more suited for a business audience.
- 1× Change the tone of this text to be more urgent and compelling. The final text must have the format <<format>>.
- 1× Clarify this text by defining any terms that might be unclear to the reader. The final text must have the topic <<topic>>.
- 1× Clarify this text by improving the phrasing while preserving its original intent. The final text must have the format <<format>> and topic <<topic>>.
- 1× Clarify this text by paraphrasing it with different vocabulary and varied sentence structures.
- 1× Clarify this text by presenting the same information from a fresh angle. The final text must have the format <<format>> and topic <<topic>>.
- 1× Edit this text to be more concise and powerful.
- 1× Edit this text to eliminate redundant and filler words. The final text must have the topic <<topic>>.
- 1× Expand this text by adding real-world applications and specific details.
- 1× Expand this text by adding storytelling elements and dialogue to increase engagement. The final text must have the format <<format>>.
- 1× Expand this text by adding supporting information to elaborate on the key points.
- 1× Expand this text by adding vivid imagery and sensory details to bring it to life.
- 1× Restructure this text to arrange the points in order of importance.
- 1× Restructure this text to eliminate awkward phrasing and improve the overall rhythm.
- 1× Restructure this text to ensure a clear introduction, body, and compelling conclusion. The final text must have the format <<format>>.
- 1× Restructure this text to group related ideas and include clear section breaks.
- 1× Restructure this text to group related ideas and include clear section breaks. The final text must have the topic <<topic>>.
- 1× Restructure this text to improve the logical flow and argument progression. The final text must have the topic <<topic>>.
- 1× Restructure this text to improve the transitions and connections between ideas.
- 1× Restructure this text to read more naturally, as if written by a native English speaker.
rewrite55 rows24 single · 31 two-turn · 37 instructions
Show the 37 instructions ▾
- 4× 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.
- 3× Edit this text by completely rewriting the topic sentence of every paragraph, while preserving the supporting sentences in those paragraphs verbatim.
- 3× Edit this text by removing all adverbs, making only minimal adjustments to the surrounding grammar to ensure the sentences remain structurally sound.
- 3× 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.
- 3× Edit this text by replacing all instances of passive voice with active voice, without altering any sentences that are already in the active voice.
- 2× 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.
- 2× Edit this text by inserting twenty to thirty words (in appropriate locations), without changing anything else.
- 2× Edit this text by removing five to ten percent of the words from it (the least necessary), without changing anything else.
- 2× Edit this text by removing one section from it (the least necessary), you may only make minimal modifications to adjacent sections for better flow.
- 2× Edit this text by replacing exactly one to three sentences, without changing anything else. The final text must have the topic <<topic>>.
- 2× Edit this text by replacing the final sentence with a more open-ended, thought-provoking statement, without altering any preceding text.
- 2× Edit this text by replacing, removing or inserting exactly one to three sentences, without changing anything else.
- 1× Edit this text by combining at least four pairs of short, adjacent sentences into complex sentences using appropriate conjunctions, leaving all other sentences untouched.
- 1× 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>>.
- 1× 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>>.
- 1× Edit this text by inserting a single, highly engaging introductory sentence at the very beginning, without modifying any of the original text that follows.
- 1× 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>>.
- 1× 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>>.
- 1× 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>>.
- 1× 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>>.
- 1× 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>>.
- 1× Edit this text by replacing exactly one to three sentences, without changing anything else.
- 1× 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>>.
- 1× Edit this text by replacing generic adjectives with highly specific, evocative ones, modifying no more than ten percent of the total words.
- 1× 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>>.
- 1× 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>>.
- 1× Edit this text by replacing, removing or inserting twenty to thirty words, without changing anything else.
- 1× Edit this text by replacing, removing or inserting twenty to thirty words, without changing anything else. The final text must have the format <<format>>.
- 1× 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>>.
- 1× 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.
- 1× 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>>.
- 1× 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>>.
- 1× 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>>.
- 1× 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>>.
- 1× 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>>.
- 1× Rewrite only the opening hook and the final concluding sentence of this text, leaving the entire middle portion exactly as it was originally written.
- 1× Rewrite the weakest section in this text, but do not make any modifications to any other section.
Topics
| Topic | Rows | Share |
|---|---|---|
| Education & Jobs | 18 | 8.2% |
| Entertainment | 16 | 7.3% |
| Finance & Business | 16 | 7.3% |
| Social Life | 14 | 6.4% |
| Software Dev. | 14 | 6.4% |
| Health | 12 | 5.5% |
| Hardware | 11 | 5.0% |
| Crime & Law | 10 | 4.5% |
| Sports & Fitness | 10 | 4.5% |
| Transportation | 10 | 4.5% |
| Travel | 9 | 4.1% |
| Games | 8 | 3.6% |
| History | 8 | 3.6% |
| Science & Tech. | 8 | 3.6% |
| Software | 8 | 3.6% |
| Food & Dining | 7 | 3.2% |
| Home & Hobbies | 7 | 3.2% |
| Industrial | 7 | 3.2% |
| Politics | 7 | 3.2% |
| Religion | 6 | 2.7% |
| Art & Design | 5 | 2.3% |
| Literature | 5 | 2.3% |
| Fashion & Beauty | 4 | 1.8% |
Formats
| Format | Rows | Share |
|---|---|---|
| Personal Blog | 48 | 21.8% |
| News Article | 30 | 13.6% |
| Nonfiction Writing | 21 | 9.5% |
| Documentation | 12 | 5.5% |
| Knowledge Article | 12 | 5.5% |
| Listicle | 12 | 5.5% |
| Customer Support | 10 | 4.5% |
| About (Pers.) | 9 | 4.1% |
| News (Org.) | 8 | 3.6% |
| Legal Notices | 7 | 3.2% |
| Q&A Forum | 7 | 3.2% |
| FAQ | 6 | 2.7% |
| Product Page | 5 | 2.3% |
| Structured Data | 5 | 2.3% |
| About (Org.) | 4 | 1.8% |
| Audio Transcript | 4 | 1.8% |
| Comment Section | 4 | 1.8% |
| Tutorial | 4 | 1.8% |
| User Review | 4 | 1.8% |
| Content Listing | 3 | 1.4% |
| Creative Writing | 3 | 1.4% |
| Academic Writing | 2 | 0.9% |
Evaluation protocol
Schema & usage73 columns · loading snippet⌄
73 columns per row. Detector columns come in pairs, one per text of the row.
Text
originalHuman-written source (class: Human)
final_responseGenerated text (class: AI)
response_0Generator output for the first turn
response_1Generator output for the second turn (empty for single-turn prompts)
Generation
promptstruct: chat_turns (list of strings), use_multiturn, examples, metadata.PROMPT_TYPE
generator_modelHugging Face id of the generator model
generation_paramsJSON string of sampling parameters
Metadata
topicTopic of the human source
formatFormat of the human source
Distances
jaccard_1, jaccard_2, levenshtein, softngram, cosdist, bertscore, bertscore_precision, bertscore_recall, moverscore, reranker10 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, original_llmjudge_valid_opus55_high, final_response_llmjudge_valid_opus55_highNot used by this report
Detector scores
One float column per text:
{original|final_response}<suffix>_editlens_score_roberta_largeEditLens Roberta-Large Score
_editlens_score_llama_3_2_3bEditLens Llama-3.2-3B Score
_editlens_score_giga_robertaGiga EditLens Roberta Score
_editlens_score_giga_llama_3_2_3bGiga EditLens Llama-3.2-3B Score
_perplexity_llama_instructPerplexity (Llama-3.2-3B-Instruct)
_perplexity_llama_basePerplexity (Llama-3.2-3B)
_entropy_llama_instructEntropy (Llama-3.2-3B-Instruct)
_entropy_llama_baseEntropy (Llama-3.2-3B)
_topp_outlier_llama_instructTop-p Outliers (Llama-3.2-3B-Instruct)
_topp_outlier_llama_baseTop-p Outliers (Llama-3.2-3B)
_topk_outlier_llama_instructTop-k Outliers (Llama-3.2-3B-Instruct)
_topk_outlier_llama_baseTop-k Outliers (Llama-3.2-3B)
_fastdetectgpt_llama_instructFastDetectGPT (Llama-3.2-3B-Instruct)
_fastdetectgpt_llama_baseFastDetectGPT (Llama-3.2-3B)
_binocularsBinoculars
_detector_score_qwen35_35b_a3bTrained detector: Qwen3.5-35B-A3B (QLoRA-style LoRA, bf16 base)
_detector_score_llama_3bTrained detector: Llama-3.2-3B (QLoRA)
_detector_score_gemma3_27bTrained detector: Gemma-3-27B (QLoRA, hard labels)
_detector_score_qwen35_35b_a3b_softTrained detector: Qwen3.5-35B-A3B soft labels (soft_ai)
_detector_score_llama_3b_softTrained detector: Llama-3.2-3B soft labels (soft_ai)
_detector_score_llama_3b_regTrained detector: Llama-3.2-3B regression (reg_sqrt)
_llmjudge_score_opus55_highLLM judge: Claude Opus 5.5 (high effort, one shot)
Load it
from datasets import concatenate_datasets, get_dataset_config_names, load_dataset
repo = "G-reen/fastdetector-test-stat-test-mini"
# 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 statistics54 statistics · all 220 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 | 0.5555 | 0.6851 | 0.2973 | 0.0262 | 0.9747 | 0 |
jaccard_2 | 0.6722 | 0.8523 | 0.3295 | 0.0343 | 0.9980 | 0 |
levenshtein | 2059.0045 | 1450.0000 | 1934.9574 | 21.0000 | 11238.0000 | 0 |
softngram | 0.5325 | 0.5833 | 0.3686 | 0.0000 | 1.0000 | 0 |
cosdist | 0.1511 | 0.0756 | 0.1788 | -0.0054 | 0.8100 | 0 |
bertscore | 0.1152 | 0.1220 | 0.0751 | 0.0020 | 0.3048 | 0 |
bertscore_precision | 0.1144 | 0.1269 | 0.0731 | 0.0022 | 0.2722 | 0 |
bertscore_recall | 0.1153 | 0.1226 | 0.0801 | 0.0016 | 0.3346 | 0 |
moverscore | 0.4734 | 0.5358 | 0.2108 | 0.0803 | 0.8534 | 0 |
reranker | -6.1672 | -7.4688 | 3.7636 | -10.0625 | 10.9375 | 0 |
EditLens Roberta-Large Score (Human) | 0.0436 | 0.0227 | 0.0568 | 0.0064 | 0.4226 | 0 |
EditLens Roberta-Large Score (AI) | 0.3323 | 0.2125 | 0.3346 | 0.0093 | 0.9996 | 0 |
EditLens Llama-3.2-3B Score (Human) | 0.0298 | 0.0147 | 0.0511 | 0.0014 | 0.4905 | 0 |
EditLens Llama-3.2-3B Score (AI) | 0.3134 | 0.2377 | 0.3019 | 0.0024 | 1.0000 | 0 |
Giga EditLens Roberta Score (Human) | 0.0091 | 0.0039 | 0.0229 | 0.0013 | 0.2062 | 0 |
Giga EditLens Roberta Score (AI) | 0.3632 | 0.1511 | 0.3952 | 0.0015 | 1.0000 | 0 |
Giga EditLens Llama-3.2-3B Score (Human) | 0.0057 | 0.0028 | 0.0155 | 0.0007 | 0.1746 | 0 |
Giga EditLens Llama-3.2-3B Score (AI) | 0.3207 | 0.1254 | 0.3665 | 0.0007 | 1.0000 | 0 |
Perplexity (Llama-3.2-3B-Instruct) (Human) | 15.2349 | 13.8759 | 6.8569 | 4.3263 | 47.1147 | 0 |
Perplexity (Llama-3.2-3B-Instruct) (AI) | 14.8530 | 13.0625 | 7.7891 | 3.4032 | 56.8310 | 0 |
Perplexity (Llama-3.2-3B) (Human) | 11.6585 | 10.8239 | 4.9460 | 1.2068 | 32.4649 | 0 |
Perplexity (Llama-3.2-3B) (AI) | 11.9417 | 10.7225 | 5.9634 | 3.0760 | 51.5935 | 0 |
Entropy (Llama-3.2-3B-Instruct) (Human) | 2.4924 | 2.4906 | 0.4558 | 1.3434 | 3.7839 | 0 |
Entropy (Llama-3.2-3B-Instruct) (AI) | 2.4182 | 2.4162 | 0.5026 | 1.0658 | 3.9175 | 0 |
Entropy (Llama-3.2-3B) (Human) | 2.3660 | 2.3909 | 0.4084 | 0.2294 | 3.4378 | 0 |
Entropy (Llama-3.2-3B) (AI) | 2.3972 | 2.3879 | 0.4125 | 1.1046 | 4.0546 | 0 |
Top-p Outliers (Llama-3.2-3B-Instruct) (Human) | 0.0537 | 0.0515 | 0.0121 | 0.0241 | 0.0942 | 0 |
Top-p Outliers (Llama-3.2-3B-Instruct) (AI) | 0.0545 | 0.0544 | 0.0120 | 0.0000 | 0.0951 | 0 |
Top-p Outliers (Llama-3.2-3B) (Human) | 0.0411 | 0.0408 | 0.0096 | 0.0074 | 0.0709 | 0 |
Top-p Outliers (Llama-3.2-3B) (AI) | 0.0391 | 0.0391 | 0.0111 | 0.0082 | 0.0805 | 0 |
Top-k Outliers (Llama-3.2-3B-Instruct) (Human) | 0.0945 | 0.0950 | 0.0350 | 0.0219 | 0.2070 | 0 |
Top-k Outliers (Llama-3.2-3B-Instruct) (AI) | 0.0915 | 0.0871 | 0.0408 | 0.0147 | 0.2546 | 0 |
Top-k Outliers (Llama-3.2-3B) (Human) | 0.0761 | 0.0748 | 0.0314 | 0.0039 | 0.1843 | 0 |
Top-k Outliers (Llama-3.2-3B) (AI) | 0.0785 | 0.0731 | 0.0376 | 0.0154 | 0.2857 | 0 |
FastDetectGPT (Llama-3.2-3B-Instruct) (Human) | -1.6189 | -1.6978 | 1.6584 | -10.5708 | 2.1883 | 0 |
FastDetectGPT (Llama-3.2-3B-Instruct) (AI) | -1.9032 | -1.9344 | 1.6282 | -7.2772 | 2.8616 | 0 |
FastDetectGPT (Llama-3.2-3B) (Human) | -0.0122 | 0.0280 | 1.0151 | -3.1103 | 2.6628 | 0 |
FastDetectGPT (Llama-3.2-3B) (AI) | 0.2631 | 0.1887 | 1.4160 | -3.6198 | 7.0747 | 0 |
Binoculars (Human) | 0.8406 | 0.8469 | 0.0793 | 0.0379 | 0.9597 | 0 |
Binoculars (AI) | 0.8621 | 0.8657 | 0.0572 | 0.4350 | 1.0098 | 0 |
Trained detector: Qwen3.5-35B-A3B (QLoRA-style LoRA, bf16 base) (Human) | 0.1128 | 0.0562 | 0.1578 | 0.0119 | 0.9999 | 0 |
Trained detector: Qwen3.5-35B-A3B (QLoRA-style LoRA, bf16 base) (AI) | 0.8465 | 1.0000 | 0.3054 | 0.0167 | 1.0000 | 0 |
Trained detector: Llama-3.2-3B (QLoRA) (Human) | 0.0599 | 0.0325 | 0.1153 | 0.0046 | 0.9607 | 0 |
Trained detector: Llama-3.2-3B (QLoRA) (AI) | 0.7707 | 0.9999 | 0.3863 | 0.0122 | 1.0000 | 0 |
Trained detector: Gemma-3-27B (QLoRA, hard labels) (Human) | 0.0699 | 0.0366 | 0.1198 | 0.0050 | 0.9050 | 0 |
Trained detector: Gemma-3-27B (QLoRA, hard labels) (AI) | 0.8224 | 1.0000 | 0.3357 | 0.0173 | 1.0000 | 0 |
Trained detector: Qwen3.5-35B-A3B soft labels (soft_ai) (Human) | 0.0373 | 0.0145 | 0.0709 | 0.0017 | 0.5713 | 0 |
Trained detector: Qwen3.5-35B-A3B soft labels (soft_ai) (AI) | 0.5547 | 0.6348 | 0.2655 | 0.0035 | 0.9399 | 0 |
Trained detector: Llama-3.2-3B soft labels (soft_ai) (Human) | 0.0467 | 0.0264 | 0.0671 | 0.0061 | 0.5122 | 0 |
Trained detector: Llama-3.2-3B soft labels (soft_ai) (AI) | 0.4886 | 0.5725 | 0.2516 | 0.0150 | 0.8808 | 0 |
Trained detector: Llama-3.2-3B regression (reg_sqrt) (Human) | 0.0114 | 0.0058 | 0.0161 | -0.0001 | 0.1187 | 0 |
Trained detector: Llama-3.2-3B regression (reg_sqrt) (AI) | 0.2745 | 0.2339 | 0.2293 | 0.0023 | 0.8516 | 0 |
LLM judge: Claude Opus 5.5 (high effort, one shot) (Human) | 0.0264 | 0.0000 | 0.0677 | 0.0000 | 0.2000 | 0 |
LLM judge: Claude Opus 5.5 (high effort, one shot) (AI) | 0.5800 | 0.8000 | 0.3286 | 0.0000 | 1.0000 | 0 |
Correlation heatmapspairwise Pearson r · Spearman ρ⌄
Each pair is correlated over the rows where both statistics are finite. Spearman compares ranks, so it shows whether a detector's score rises with edit distance even when the relationship is not linear.
Pearson r
Spearman ρ
Model-specific analyticsseparate page · 22 detectors⌄
Per-detector score grids by topic × format and by generator × prompt category, and the ranked specific-prompt tables, are on their own page to keep this card within the Hub's size limit: open ANALYTICS.md ↗.
Generated by the FastDetector analysis pipeline from the complete published corpus.config/analysis_mini.toml
- Downloads last month
- 1,115