open text detector: research mixes
Collection
training mixes from runs that were compared but not used by the model on https://opentextdetector.com • 2 items • Updated
id stringlengths 11 78 | split_file stringclasses 1
value | dataset stringclasses 4
values | kind stringclasses 3
values | label stringclasses 3
values | ai_chars int32 0 3.4k | text stringlengths 250 3.4k | regions listlengths 1 3 | document_label int8 | supervision stringclasses 1
value | generator stringclasses 3
values | source stringclasses 34
values | venue stringclasses 5
values | year int16 2.01k 2.03k ⌀ | text_format stringclasses 0
values | paper_id stringlengths 15 70 | group stringlengths 21 70 | host_id stringclasses 0
values | seed_id stringclasses 0
values | source_start int32 0 46.3k | source_end int32 250 48.8k | target_start int32 0 2.25k | target_end int32 0 3.4k | original_split stringclasses 0
values | protected bool 0
classes | draw_id stringlengths 15 19 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
luna-mirror/08d5a2f76432d1b32d17419d484d4da01d87cdce42aab5fc6d001b61e884c268@0 | stage1-epoch0 | mirrors | generated | ai | 2,954 | , for example, an entity’s employer from its location. Stacking layers incorporates progressively wider neighborhoods, while normalization or residual connections can help preserve informative textual features. The resulting representation reflects both what an entity’s description says and how it is situated in the ob... | [
{
"start": 0,
"end": 2954,
"label": 1
}
] | null | spans | openai/gpt-6-luna | acl | null | null | null | human:6bb65ed17b5ce0c81713bf6bba82a8e335b017d23639c49c6094326d94c92302 | human:6bb65ed17b5ce0c81713bf6bba82a8e335b017d23639c49c6094326d94c92302 | null | null | 1,809 | 4,763 | 0 | 2,954 | null | null | stage1-epoch0-7308 |
a3e2206f39790dc59c013bc44e2dab737bfc21fc30917b035294a18b16af280d | stage1-epoch0 | human | novel_human | human | 0 | Just returned from a week at the Homewood in Chicago. Lived up pto what I expect from Homewoods. Great breakfast and cocktail/snacks hour. Lots of kids b/c of spring break. Not too bad though. Great location. Only complaint - elevators take a while, especially when one is down for a couple of days. | [
{
"start": 0,
"end": 299,
"label": 0
}
] | null | spans | null | opinrank | null | null | null | human:f7a430b8535c9e01937ac8eb8f9bfc95625a1c1775a69cc7285502bf62d9098b | human:f7a430b8535c9e01937ac8eb8f9bfc95625a1c1775a69cc7285502bf62d9098b | null | null | 0 | 299 | 0 | 299 | null | null | stage1-epoch0-1622 |
f90d583af1ead4aa753be0c972d2c41308b6e12f5731b48ec9a19cc49bdd76fc | stage1-epoch0 | human | novel_human | human | 0 | in central Ethiopia threatens farmers' incomes and public health, according to a study.
As a result, the study calls for a concerted effort to control the disease and prevent the transmission of Mycobacterium bovis, the bacteria that causes it, to humans consuming dairy products from these farms.
SPEED READ
• ... | [
{
"start": 0,
"end": 2507,
"label": 0
}
] | null | spans | null | scidev | null | null | null | human:9b855b677bd76518056491d5ea9dad1e9f56669b7e8e3c9c72b625ccb8baddac | human:9b855b677bd76518056491d5ea9dad1e9f56669b7e8e3c9c72b625ccb8baddac | null | null | 74 | 2,581 | 0 | 2,507 | null | null | stage1-epoch0-6602 |
icml-2015-c445f519f82cbf58/g04/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 749 | Baselines and method. We compare HashedNets with several existing techniques for size-constrained, feedforward neural networks. Random Edge Removal (RER) (Cireşan et al., 2011) reduces the total number of model parameters by randomly removing weights prior to training. Low-Rank Decomposition (LRD) (Denil et al., 2013) ... | [
{
"start": 0,
"end": 597,
"label": 0
},
{
"start": 597,
"end": 1346,
"label": 1
},
{
"start": 1346,
"end": 1778,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ICML | 2,015 | null | icml-2015-c445f519f82cbf58 | paper:icml-2015-c445f519f82cbf58 | null | null | 0 | 1,778 | 597 | 1,346 | null | null | stage1-epoch0-6572 |
a0afdf1ac166b8652ffe9dee6eac779e/g04/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 835 | contextual clues. Examples of such processes are common, e.g. visual illusions like the Necker cube [38], and may be driven by the functional organisation and the heavy top-down influences within the ventral visual stream of the brain [19, 41]. To evaluate the ability of VASE to deal with ambiguous inputs based on the... | [
{
"start": 0,
"end": 1390,
"label": 0
},
{
"start": 1390,
"end": 2225,
"label": 1
},
{
"start": 2225,
"end": 2530,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | NeurIPS | 2,018 | null | a0afdf1ac166b8652ffe9dee6eac779e | paper:a0afdf1ac166b8652ffe9dee6eac779e | null | null | 102 | 2,632 | 1,390 | 2,225 | null | null | stage1-epoch0-1252 |
icml-2021-df3a234d23bd34f2/g05/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 472 | networks do not come with robustness guarantees. To address this, different certification methods have been proposed recently using various methods, relying upon SMT solvers (Katz et al., 2017; Ehlers, 2017), semidefinite programming (Raghunathan et al., 2018a) and linear relaxations (Gehr et al., 2018; Zhang et al., ... | [
{
"start": 0,
"end": 660,
"label": 0
},
{
"start": 660,
"end": 1132,
"label": 1
},
{
"start": 1132,
"end": 1889,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ICML | 2,021 | null | icml-2021-df3a234d23bd34f2 | paper:icml-2021-df3a234d23bd34f2 | null | null | 71 | 1,960 | 660 | 1,132 | null | null | stage1-epoch0-5353 |
30dbb12d765796efc3c23c70693843890410a725b149cdbcc2245f796c2b6196 | stage1-epoch0 | human | novel_human | human | 0 | Implications for Forest management
From the analysis of the number of individuals of each species, it appears that some species (12) are very poorly represented in the community forest (density less than 0.03 stems ha À1 ). In order to preserve them, we recommend banning their exploitation. Among these species are firs... | [
{
"start": 0,
"end": 1057,
"label": 0
}
] | null | spans | null | pes2o | null | null | null | human:0e22ce6599620204bd75019f6c6390e2b3d284fae7ef13f90ec35cd693ed950f | human:0e22ce6599620204bd75019f6c6390e2b3d284fae7ef13f90ec35cd693ed950f | null | null | 0 | 1,057 | 0 | 1,057 | null | null | stage1-epoch0-3680 |
luna-mirror/aaa21d15e139e5dd2970356ca810f8549820793ed7b1aa253c4da837f0c33021 | stage1-epoch0 | mirrors | generated | ai | 2,738 | , often blend comedy, social observation, and psychological depth.
The term “classical opera” can refer broadly to the Western operatic tradition, but it can also suggest music of the Classical period, roughly the later eighteenth century. That era favored balance, clarity, and elegant musical forms. In the nineteenth... | [
{
"start": 0,
"end": 2738,
"label": 1
}
] | null | spans | openai/gpt-6-luna | libretexts | null | null | null | human:0543584e68160a093580f8e725c7f17318a606cdda44c5a1575508d5e5ece456 | human:0543584e68160a093580f8e725c7f17318a606cdda44c5a1575508d5e5ece456 | null | null | 2,989 | 5,727 | 0 | 2,738 | null | null | stage1-epoch0-1309 |
luna-mirror/14259753c60eea21fbee78c60ba2164a3bbb7537fcacd07139dffc57affcbf0c@0 | stage1-epoch0 | mirrors | generated | ai | 2,631 | valuable even without covering every part of the program. Teams needed women and men who could handle singles, reliable doubles pairings, and players comfortable switching partners in mixed doubles.
The Freedoms’ roster decisions reflected that structure. The team relied on experienced doubles players, notably Lisa R... | [
{
"start": 0,
"end": 2631,
"label": 1
}
] | null | spans | openai/gpt-6-luna | wikipedia | null | null | null | human:03da4b95082e35285de75042d98667e19c0a03471a6ff19e2694871c32b04fad | human:03da4b95082e35285de75042d98667e19c0a03471a6ff19e2694871c32b04fad | null | null | 962 | 3,593 | 0 | 2,631 | null | null | stage1-epoch0-4405 |
luna-mirror/a4bd4211f55d06145b944629148df792fc83836301c19f2617120b3e1feeaaac | stage1-epoch0 | mirrors | generated | ai | 2,168 | **Clinton’s Middle East Message Meets Washington’s Impeachment Storm**
President Bill Clinton’s latest remarks on the Middle East drew attention this week not only for their emphasis on diplomacy, but for the political crisis unfolding at home. He urged Israelis and Palestinians to keep negotiating, arguing that viole... | [
{
"start": 0,
"end": 2168,
"label": 1
}
] | null | spans | openai/gpt-6-luna | oanc_slate | null | null | null | human:4b6881481a9ed39dda5280f04f3630211a23a85bf80b793a7f684efac9b5b65d | human:4b6881481a9ed39dda5280f04f3630211a23a85bf80b793a7f684efac9b5b65d | null | null | 0 | 2,168 | 0 | 2,168 | null | null | stage1-epoch0-2027 |
79c8a30c4f0ddbc4e4aa126f887f86219440e444706d999f3799cac2eaaa57ac | stage1-epoch0 | human | novel_human | human | 0 | Truss the birds, putting aside the hearts, livers, and gizzards, and
dredge them with flour, then place them in a saucepan with a piece of
butter, and let them brown equally, taking care of the gravy which oozes
from them. Let them get cold, then carve them in such a way that the
wings and legs can be taken off with a ... | [
{
"start": 0,
"end": 924,
"label": 0
}
] | null | spans | null | gutenberg | null | null | null | human:e773dd3fffa2f6cc0fd3866762ae4ef7be0364cd146dc405300c4c891bc1549a | human:e773dd3fffa2f6cc0fd3866762ae4ef7be0364cd146dc405300c4c891bc1549a | null | null | 0 | 924 | 0 | 924 | null | null | stage1-epoch0-8761 |
fullpaper-252@0 | stage1-epoch0 | fullpapers | generated | ai | 2,153 | mathrm{span}}+\lambda_c\mathcal L_{\mathrm{cons}},
$$
where $q_t$ predicts observed prefix violation and $s_{i\mid t}$ predicts token-span membership for $i\leq t$ using only prefix $y_{1:t}$. The span term is the masked, duplicate-weighted average of $\operatorname{BCE}(s_{i\mid t},a_i)$ over available prefix tokens,... | [
{
"start": 0,
"end": 2153,
"label": 1
}
] | null | spans | openai/gpt-6-sol | openreview | NeurIPS | 2,025 | null | neurips:mnIox70nhO | paper:neurips:mnIox70nhO | null | null | 11,237 | 13,390 | 0 | 2,153 | null | null | stage1-epoch0-899 |
acl-2021-8929038991115d00/g03/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 500 | We recruited 3 annotators, from a pool of internship students, as reviewers over a period of 18 weeks to filter and post-edit the generated pairs after an extensive training procedure.
Training. The annotators underwent 2 weeks of training, which made them “experts” in HS/CN post-editing. The training comprised readin... | [
{
"start": 0,
"end": 186,
"label": 0
},
{
"start": 186,
"end": 686,
"label": 1
},
{
"start": 686,
"end": 1497,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ACL | 2,021 | null | acl-2021-8929038991115d00 | paper:acl-2021-8929038991115d00 | null | null | 0 | 1,497 | 186 | 686 | null | null | stage1-epoch0-1976 |
1bd4b29a8e0afccd9923fe29cecb4b29/g04/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 1,608 | respectively, and their joint effect improves retrieval performance significantly. Our model agrees with the push-pull phenomenon observed in neural data and sheds light on our understanding of the role of feedback in neural information processing.
Deep neural networks (DNNs) mimic hierarchical processing in the vent... | [
{
"start": 0,
"end": 251,
"label": 0
},
{
"start": 251,
"end": 1859,
"label": 1
},
{
"start": 1859,
"end": 2898,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | NeurIPS | 2,019 | null | 1bd4b29a8e0afccd9923fe29cecb4b29 | paper:1bd4b29a8e0afccd9923fe29cecb4b29 | null | null | 786 | 3,684 | 251 | 1,859 | null | null | stage1-epoch0-10575 |
6e2713a6efee97bacb63e52c54f0ada0/g03/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 1,183 | The newly proposed deep neural decision forests (dNDFs) [20] overcomes this problem by introducing a soft differentiable decision function at the split nodes and a global loss function defined on a tree. This ensures that the split node parameters can be learned by back-propagation and leaf node predictions can be upda... | [
{
"start": 0,
"end": 359,
"label": 0
},
{
"start": 359,
"end": 1542,
"label": 1
},
{
"start": 1542,
"end": 2124,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | NeurIPS | 2,017 | null | 6e2713a6efee97bacb63e52c54f0ada0 | paper:6e2713a6efee97bacb63e52c54f0ada0 | null | null | 0 | 2,124 | 359 | 1,542 | null | null | stage1-epoch0-10931 |
luna-mirror/3683eca31d15c157abf0445ecff26c6dddf59d2aa0e9a8d9f3fb329de3ced459 | stage1-epoch0 | mirrors | generated | ai | 3,062 | weighing the country’s choices and challenges from several angles, with recent posts ranging from travel advice to debates over education, politics and cultural change. Together, the writing offers a lively picture of how everyday concerns meet larger national conversations online.
Tourism posts look beyond familiar ... | [
{
"start": 0,
"end": 3062,
"label": 1
}
] | null | spans | openai/gpt-6-luna | globalvoices | null | null | null | human:bcf1259c511fb78fbc07a35bc2f56a92502d5eae86c3bb0351a127420bc36671 | human:bcf1259c511fb78fbc07a35bc2f56a92502d5eae86c3bb0351a127420bc36671 | null | null | 24 | 3,086 | 0 | 3,062 | null | null | stage1-epoch0-786 |
luna-mirror/ef8e76d3e5d80ff0845932fdffa2886b0848ce48b521cb8f25f7583edcbcaedb | stage1-epoch0 | mirrors | generated | ai | 1,216 | I booked three nights at the Harbor House because its photos promised a quiet, polished weekend getaway. The room was clean and the bed comfortable, though the bathroom fixtures looked tired and the hallway carried a smell of cooking oil. Check-in was smooth, and staff were courteous.
My main frustration was the const... | [
{
"start": 0,
"end": 1216,
"label": 1
}
] | null | spans | openai/gpt-6-luna | opinrank | null | null | null | human:4306c74c108c1561a7ebede2453215b783da4eea186389bb38bfa4f697941483 | human:4306c74c108c1561a7ebede2453215b783da4eea186389bb38bfa4f697941483 | null | null | 0 | 1,216 | 0 | 1,216 | null | null | stage1-epoch0-3608 |
luna-mirror/9c9ccb0c59286d1f29fe374d0f30d4b2ef6438dc0be832ad3c4b87e1c764059b | stage1-epoch0 | mirrors | generated | ai | 2,359 | Phillips acknowledged her and gave the position again. His fingers moved without hesitation. Bride watched the message take shape, dot by dot, dash by dash, and thought of the thousands of miles of water between the two ships. A call could cross that distance. A lifeboat could not.
The captain appeared briefly. He gav... | [
{
"start": 0,
"end": 2359,
"label": 1
}
] | null | spans | openai/gpt-6-luna | gutenberg | null | null | null | human:d06bb87956c60cd63080863c32f064e9e848f0671670398176d738fad5a128ea | human:d06bb87956c60cd63080863c32f064e9e848f0671670398176d738fad5a128ea | null | null | 2,177 | 4,536 | 0 | 2,359 | null | null | stage1-epoch0-4445 |
luna-mirror/b4cefcddc07904b332f419cbeb5486b120202b590059c18eaaf9ca233ccae484@0 | stage1-epoch0 | mirrors | generated | ai | 2,979 | STES) addresses this challenge by shifting heat across time. Summer heat rejected during cooling can be stored for winter heating, while winter heat extracted from the ground can, in some applications, support summer cooling. Borehole fields can thus function as both heat-exchange infrastructure and distributed storage... | [
{
"start": 0,
"end": 2979,
"label": 1
}
] | null | spans | openai/gpt-6-luna | pes2o | null | null | null | human:18288fb13bae71abbbef2de3b52832e1496fe2e830d0122d276715b774f12df8 | human:18288fb13bae71abbbef2de3b52832e1496fe2e830d0122d276715b774f12df8 | null | null | 4,452 | 7,431 | 0 | 2,979 | null | null | stage1-epoch0-1586 |
acl-2013-c6cc7c36cf0c99d1/g02/human_original | stage1-epoch0 | papers | paired | human | 0 | In contrast to TINE, MEANT (Lo et al., 2012), which is the weighted f-score over the matched semantic role labels of the automatically aligned semantic frames and role fillers, outperforms BLEU, NIST, METEOR, WER, CDER and TER. This makes it more suitable for tuning SMT systems to produce much adequate translations.
A... | [
{
"start": 0,
"end": 878,
"label": 0
}
] | null | spans | null | proceedings | ACL | 2,013 | null | acl-2013-c6cc7c36cf0c99d1 | paper:acl-2013-c6cc7c36cf0c99d1 | null | null | 0 | 878 | 319 | 614 | null | null | stage1-epoch0-6796 |
3341f6f048384ec73a7ba2e77d2db48b/g02/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 730 | between the target and source domain. It violates the basic assumption of conventional machine learning algorithms that training samples and test samples sharing the same distribution. To bridge the domain gap, UDA approaches match either the marginal distributions [30, 21, 11, 33] or the conditional distributions [38... | [
{
"start": 0,
"end": 787,
"label": 0
},
{
"start": 787,
"end": 1517,
"label": 1
},
{
"start": 1517,
"end": 2342,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | NeurIPS | 2,019 | null | 3341f6f048384ec73a7ba2e77d2db48b | paper:3341f6f048384ec73a7ba2e77d2db48b | null | null | 78 | 2,420 | 787 | 1,517 | null | null | stage1-epoch0-10198 |
luna-mirror/32af128558acee1d17913a761fb1fa4738c49f3fce1e5531dd842c5b7e5f0d78 | stage1-epoch0 | mirrors | generated | ai | 2,684 | OERlabs can be understood as a practical inquiry into how open educational resources (OER) and open educational practices (OEP) might become part of teacher education, rather than remain abstract policy goals. The project treats student teachers not merely as recipients of materials but as participants who investigate ... | [
{
"start": 0,
"end": 2684,
"label": 1
}
] | null | spans | openai/gpt-6-luna | wikibooks | null | null | null | human:575cdf5c83181c3bf6c1604c8726dcf74a5270c509e3e241ff5f00bce70b7539 | human:575cdf5c83181c3bf6c1604c8726dcf74a5270c509e3e241ff5f00bce70b7539 | null | null | 0 | 2,684 | 0 | 2,684 | null | null | stage1-epoch0-9829 |
acl-2020-b85e3d7dc455646e/g05/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 697 | The dodecaDialogue task is intended to assemble important aspects of an engaging conversational agent into a single collection, where each subtask covers some of those goals. Such an agent should be able to get to know you when you first talk to it (ConvAI2), discuss everyday topics (DailyDialog, pushshift.io Reddit, T... | [
{
"start": 0,
"end": 699,
"label": 0
},
{
"start": 699,
"end": 1396,
"label": 1
},
{
"start": 1396,
"end": 1974,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ACL | 2,020 | null | acl-2020-b85e3d7dc455646e | paper:acl-2020-b85e3d7dc455646e | null | null | 0 | 1,974 | 699 | 1,396 | null | null | stage1-epoch0-2010 |
icml-2015-a8e5d856487043c4/g03/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 1,004 | Significance of our experiments is that we apply SUMO to process three real world large scale corpora using only 3GB of RAM, which is beyond the scope of the existing Monte Carlo methods. This marks a significant improvement in Monte Carlo family for learning with large datasets. SUMO makes it possible as memory usage ... | [
{
"start": 0,
"end": 382,
"label": 0
},
{
"start": 382,
"end": 1386,
"label": 1
},
{
"start": 1386,
"end": 1859,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ICML | 2,015 | null | icml-2015-a8e5d856487043c4 | paper:icml-2015-a8e5d856487043c4 | null | null | 0 | 1,859 | 382 | 1,386 | null | null | stage1-epoch0-10158 |
39027dfad5138c9ca0c474d71db915c3/g04/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 636 | Learning from historical logs data can be formalized as batch learning from logged bandit feedback (BLBF) [2, 1]. Unlike the well-studied problem of online learning from bandit feedback [3], this setting does not require the learner to have interactive control over the system. Learning in such a setting is closely rela... | [
{
"start": 0,
"end": 758,
"label": 0
},
{
"start": 758,
"end": 1394,
"label": 1
},
{
"start": 1394,
"end": 1982,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | NeurIPS | 2,015 | null | 39027dfad5138c9ca0c474d71db915c3 | paper:39027dfad5138c9ca0c474d71db915c3 | null | null | 0 | 1,982 | 758 | 1,394 | null | null | stage1-epoch0-6856 |
luna-mirror/16e1bc94263126497ffde44a53a352ca617c5fd716ba75d479e2f937615c6c4f | stage1-epoch0 | mirrors | generated | ai | 1,766 | Antibody distribution within solid tumors is shaped not only by molecular properties such as size and affinity, but also by the physiological state of tumor blood vessels. Unlike well-organized vasculature in healthy tissues, tumor vessels are often irregularly branched, structurally immature, and variably perfused. Th... | [
{
"start": 0,
"end": 1766,
"label": 1
}
] | null | spans | openai/gpt-6-luna | pmc | null | null | null | human:f1361ca84261f0015344420cb01fec14daa4ed50b531fd6ba4682ab8c93c726c | human:f1361ca84261f0015344420cb01fec14daa4ed50b531fd6ba4682ab8c93c726c | null | null | 0 | 1,766 | 0 | 1,766 | null | null | stage1-epoch0-4185 |
luna-mirror/82d4e2a37b3eba543d208f0740d7c1d5a660b28d8d0515053bfb6c1d13372201 | stage1-epoch0 | mirrors | generated | ai | 1,433 | I picked up The Glass Orchard and was won over by its strange atmosphere and messy, believable characters. Mara, a practical mapmaker who distrusts magic, is especially engaging; her humor hides a grief that the book reveals gradually rather than explaining all at once. Her uneasy partnership with the exiled prince, Iv... | [
{
"start": 0,
"end": 1433,
"label": 1
}
] | null | spans | openai/gpt-6-luna | amazon2018 | null | null | null | human:f811b89759d3582c743aadf30c16dce75fd0ade52622fe52d901aefa22d21500 | human:f811b89759d3582c743aadf30c16dce75fd0ade52622fe52d901aefa22d21500 | null | null | 0 | 1,433 | 0 | 1,433 | null | null | stage1-epoch0-6526 |
acl-2019-3c8feabdf8f13efc/g05/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 460 | Starting with our annotated dataset, which includes miniclips paired with transcripts and candidate actions drawn from the transcript, we extract several layers of information, which we then use to develop our multimodal model, as well as several baselines.
**Action Embeddings.** We use GloVe (Pennington et al., 2014)... | [
{
"start": 0,
"end": 259,
"label": 0
},
{
"start": 259,
"end": 719,
"label": 1
},
{
"start": 719,
"end": 1227,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ACL | 2,019 | null | acl-2019-3c8feabdf8f13efc | paper:acl-2019-3c8feabdf8f13efc | null | null | 0 | 1,227 | 259 | 719 | null | null | stage1-epoch0-5593 |
luna-mirror/22c64b2475f16d575dbad4b611ecebde19661f47045674ea068097d6d8f8a23a | stage1-epoch0 | mirrors | generated | ai | 2,330 | Reflective writing helps writers understand not only what they produce but how they produce it. Metacognition, or thinking about one’s thinking, gives writers a way to notice choices, habits, and assumptions while composing. Instead of treating a draft as a finished object, reflection turns it into evidence: a record o... | [
{
"start": 0,
"end": 2330,
"label": 1
}
] | null | spans | openai/gpt-6-luna | libretexts | null | null | null | human:152c63bcc089059df8dd054b39c0a921807224fad07b79e2cacd63eba8d286ea | human:152c63bcc089059df8dd054b39c0a921807224fad07b79e2cacd63eba8d286ea | null | null | 0 | 2,330 | 0 | 2,330 | null | null | stage1-epoch0-9468 |
acl-2020-98e598de672129f9/g01/human_original | stage1-epoch0 | papers | paired | human | 0 | When humans are tasked with constructing a time-line, we expect that they reason over important events rather than dates. Conceptually, detecting and selecting events might also be more appropriate than selecting dates because multiple events can happen on the same day, and an event can potentially span multiple days. ... | [
{
"start": 0,
"end": 1851,
"label": 0
}
] | null | spans | null | proceedings | ACL | 2,020 | null | acl-2020-98e598de672129f9 | paper:acl-2020-98e598de672129f9 | null | null | 0 | 1,851 | 705 | 1,428 | null | null | stage1-epoch0-6493 |
acl-2016-aef39440c7ac2622/g02/human_original | stage1-epoch0 | papers | paired | human | 0 | The Parallel-Hierarchical model is simple. It does no complex language or sequence modeling. Its simplicity is a response to the limited data of MCTest. Nevertheless, the model achieves stateof-the-art results on the multi questions, which (putatively) require some limited reasoning. Our model is able to handle them re... | [
{
"start": 0,
"end": 1768,
"label": 0
}
] | null | spans | null | proceedings | ACL | 2,016 | null | acl-2016-aef39440c7ac2622 | paper:acl-2016-aef39440c7ac2622 | null | null | 0 | 1,768 | 574 | 1,528 | null | null | stage1-epoch0-6837 |
acl-2019-3138e2ffe0923c51/g02/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 1,441 | optimal parameter choices remain unclear, because up to now most models have been compared on different evaluation tasks and data. Presently, we do not know which model performs best under which conditions, and if more complex model architectures gain performance benefits over simpler models. This situation hinders ad... | [
{
"start": 0,
"end": 537,
"label": 0
},
{
"start": 537,
"end": 1978,
"label": 1
}
] | null | spans | openai/gpt-6-luna | proceedings | ACL | 2,019 | null | acl-2019-3138e2ffe0923c51 | paper:acl-2019-3138e2ffe0923c51 | null | null | 310 | 2,288 | 537 | 1,978 | null | null | stage1-epoch0-1836 |
acl-2016-cda98c0a5a00632d/g01/human_original | stage1-epoch0 | papers | paired | human | 0 | 13). We investigate if adding depth to our context-sensitive autoencoder will improve its performance in the contextual word similarity task. Figure 6 shows that as we increase the depth of our autoencoders, their performances initially improve. The CSAE-LGC model that uses both local and global context benefits more f... | [
{
"start": 0,
"end": 2191,
"label": 0
}
] | null | spans | null | proceedings | ACL | 2,016 | null | acl-2016-cda98c0a5a00632d | paper:acl-2016-cda98c0a5a00632d | null | null | 220 | 2,411 | 647 | 1,674 | null | null | stage1-epoch0-2032 |
c47d5dca76c180aa7767e0565b413935e4be50065fad5836e40f8e47be35aa5a | stage1-epoch0 | human | novel_human | human | 0 | thoughts , with her red hair whipping back , belly slightly swollen with another child… <newline> <newline> He smiled and dismissed those thoughts , those happy memories . Now wasn ’ t the time for reminiscing , he had a mission this time . The dark wizard he was after was a fairly new problem , but he had caused a lo... | [
{
"start": 0,
"end": 2322,
"label": 0
}
] | null | spans | null | writingprompts | null | null | null | human:2bce83e237c4b5dc62fed528a6ff44d683828074b66271b9cf43bdd744852894 | human:2bce83e237c4b5dc62fed528a6ff44d683828074b66271b9cf43bdd744852894 | null | null | 1,418 | 3,740 | 0 | 2,322 | null | null | stage1-epoch0-7942 |
df72d973d5fd632448761ba2a8301b30c1580f83b2246c217b37452aa5535095 | stage1-epoch0 | human | novel_human | human | 0 | 30 min and centrifuged at 14,000 ×*g* for 10 min at 4 °C. The supernatants were transferred to fresh tubes, and the pellets were mixed with cold IP lysis buffer before sonication, followed by a second round of centrifugation at 14,000 ×*g* for 10 min at 4 °C. The supernatants obtained from the two extraction steps were... | [
{
"start": 0,
"end": 1499,
"label": 0
}
] | null | spans | null | pmc | null | null | null | human:ee0e7ba9aee17094c4345692fedd7c50b02f22bdce4a78bec41ee87aa58dd538 | human:ee0e7ba9aee17094c4345692fedd7c50b02f22bdce4a78bec41ee87aa58dd538 | null | null | 594 | 2,093 | 0 | 1,499 | null | null | stage1-epoch0-4983 |
icml-2016-2ccc43de45d13b0a/g05/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 559 | : P (X) = Z 1 i ψ i (X i ), where ψ i is a potential over variables X i ⊆ X and Z is known as the partition function (Pearl, 1988). One of the main inference problems in graphical models is to compute the probability P of evidence e ∈ X E for variables X E ⊆ X, P (e) = X P (e, X E ), where X E = X\X E . The par- E titi... | [
{
"start": 0,
"end": 1158,
"label": 0
},
{
"start": 1158,
"end": 1717,
"label": 1
},
{
"start": 1717,
"end": 1912,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ICML | 2,016 | null | icml-2016-2ccc43de45d13b0a | paper:icml-2016-2ccc43de45d13b0a | null | null | 81 | 1,993 | 1,158 | 1,717 | null | null | stage1-epoch0-11910 |
icml-2020-36249ac0752ff815/g04/human_original | stage1-epoch0 | papers | paired | human | 0 | , the local worst-case risk is defined as the supremum of the risk in the vicinity of the empirical data distribution, called the ambiguity set. The ambiguity set is often designed as a neighborhood of P n and the closeness of two measures is evaluated by φ-divergences or probability metrics. Note that WDRO is a specia... | [
{
"start": 0,
"end": 1830,
"label": 0
}
] | null | spans | null | proceedings | ICML | 2,020 | null | icml-2020-36249ac0752ff815 | paper:icml-2020-36249ac0752ff815 | null | null | 128 | 1,958 | 668 | 1,363 | null | null | stage1-epoch0-4856 |
icml-2016-f1a7099d524e5411/g03/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 390 | The middle row of Fig. 7 shows results from the lossy compression task. We use the sieve to transform the original digits into 12 binary latent factors, Y , plus remainder in- 12 formation for each pixel, X 1:784 , and then we use the Y ’s alone to reconstruct the image. In the third row, the Y ’s are estimated using o... | [
{
"start": 0,
"end": 531,
"label": 0
},
{
"start": 531,
"end": 921,
"label": 1
},
{
"start": 921,
"end": 1625,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ICML | 2,016 | null | icml-2016-f1a7099d524e5411 | paper:icml-2016-f1a7099d524e5411 | null | null | 0 | 1,625 | 531 | 921 | null | null | stage1-epoch0-5998 |
9b7ac2ebbe7c35ed97dd2d99966bf614b88196ee7b6774a4a56eccc8a43be6ce | stage1-epoch0 | human | novel_human | human | 0 | That night and the next few days are hazy in the memory for obvious
reasons. It was a risk to go back to my room, but a calculated one.
The chances were good that Angelina didn't know of its existence or,
even if she had found out, that she wouldn't have done anything about
it. I was dead and she had no further interes... | [
{
"start": 0,
"end": 1321,
"label": 0
}
] | null | spans | null | gutenberg | null | null | null | human:90736cf75c95fcca56cdb9f8fbae979e9a0761557f871f48af6a96e88d920ba5 | human:90736cf75c95fcca56cdb9f8fbae979e9a0761557f871f48af6a96e88d920ba5 | null | null | 0 | 1,321 | 0 | 1,321 | null | null | stage1-epoch0-6823 |
acl-2015-9a5f39fa7b1afe20/g01/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 588 | We have built the largest Tibetan text resources over the internet via an automatic crawler. They are from three web sites, that are, Tibet Daily, People’s Daily and Qinghai Daily. This News Corpus includes different fields such as politics, science, technology, education, language and culture, religion, tourism, envir... | [
{
"start": 0,
"end": 740,
"label": 0
},
{
"start": 740,
"end": 1328,
"label": 1
},
{
"start": 1328,
"end": 1649,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ACL | 2,015 | null | acl-2015-9a5f39fa7b1afe20 | paper:acl-2015-9a5f39fa7b1afe20 | null | null | 0 | 1,649 | 740 | 1,328 | null | null | stage1-epoch0-3192 |
2f2b265625d76a6704b08093c652fd79/g04/human_original@87 | stage1-epoch0 | papers | paired | human | 0 | Comparing the accuracy gains of different methods after applying CCVR and whole data calibration, we find that the accuracies of FedAvg and MOON get the greatest increase. On CINIC-10, the oracle results of FedAvg even outstrip those of all other baselines, implying that FedAvg focuses more on learning high-quality fea... | [
{
"start": 0,
"end": 2453,
"label": 0
}
] | null | spans | null | proceedings | NeurIPS | 2,021 | null | 2f2b265625d76a6704b08093c652fd79 | paper:2f2b265625d76a6704b08093c652fd79 | null | null | 427 | 2,880 | 428 | 1,181 | null | null | stage1-epoch0-9157 |
95a6fc111fa11c3ab209a0ed1b9abeb6/g01/human_original | stage1-epoch0 | papers | paired | human | 0 | CPU Run Time. We show the CPU run time plots of WERM-ID-R versus Plug-in over increasing D in Fig. 3d, 3e, 3f. The run time plots of WERM-ID always overlap with WERM-ID-R and are therefore not shown. For each given D, we collect 100 run times and plot the median. We note that, in all three experiments, the run time of ... | [
{
"start": 0,
"end": 2222,
"label": 0
}
] | null | spans | null | proceedings | NeurIPS | 2,020 | null | 95a6fc111fa11c3ab209a0ed1b9abeb6 | paper:95a6fc111fa11c3ab209a0ed1b9abeb6 | null | null | 0 | 2,222 | 429 | 1,351 | null | null | stage1-epoch0-10436 |
acl-2019-a8919de3045564e3/g04/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 278 | analyze language shift. The Law of Conformity states a negative correlation between word frequency and meaning change (Hamilton et al., 2016); however, Dubossarsky et al. (2017) show that the word frequency does play an important role in the semantic change, even though a small one. Diachronic embeddings have been use... | [
{
"start": 0,
"end": 1644,
"label": 0
},
{
"start": 1644,
"end": 1922,
"label": 1
},
{
"start": 1922,
"end": 2383,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ACL | 2,019 | null | acl-2019-a8919de3045564e3 | paper:acl-2019-a8919de3045564e3 | null | null | 123 | 2,506 | 1,644 | 1,922 | null | null | stage1-epoch0-10136 |
8993a9cc0c6863ece169ff000f630d61c28ccd47d6b46a3ef3fc4829a8f846a4 | stage1-epoch0 | human | novel_human | human | 0 | of the Republican party is not just now
the revival of the African slave-trade, or the passage of a
Congressional slave-code ... but the most imminent danger that now
threatens that purpose is that insidious Douglas popular sovereignty.
This is the miner and sapper. While it does not propose to revive the
African slav... | [
{
"start": 0,
"end": 2336,
"label": 0
}
] | null | spans | null | gutenberg | null | null | null | human:0c703f1a848999fcd6ab54bba64bb8412a2c4f9b1e31ebf774114c11a6c5cd22 | human:0c703f1a848999fcd6ab54bba64bb8412a2c4f9b1e31ebf774114c11a6c5cd22 | null | null | 32 | 2,368 | 0 | 2,336 | null | null | stage1-epoch0-9201 |
3974fb04b1463955f1f91668cedf4c5b7822c4d8f4436f1b5cecd6f3b0c370a9 | stage1-epoch0 | human | novel_human | human | 0 | Among the quivering needles
the fly
smashes, its imp-wings
glimmering in the cooling light, its demon eyes
shattered into broken mirrors
replicating ad infinitum car car car car
tree tree tree ... now it lands on the steering wheel
and faces me:
What do I look like
in the eyes of a fly?
... | [
{
"start": 0,
"end": 751,
"label": 0
}
] | null | spans | null | oanc_slate | null | null | null | human:1171e9565ea1aae5ac7c0388d8d86a8e933dcd07cd3d6ed3e37663f954a9a248 | human:1171e9565ea1aae5ac7c0388d8d86a8e933dcd07cd3d6ed3e37663f954a9a248 | null | null | 0 | 751 | 0 | 751 | null | null | stage1-epoch0-10184 |
acl-2019-911d23b62de2a4e0/g05/human_original | stage1-epoch0 | papers | paired | human | 0 | Tribal affiliation Nigeria is home to individuals identifying with over a hundred different tribal identities which are concentrated in different regions. These tribal affiliations are the strongest aspect of self identity in present day Nigeria (Mustapha, 2006) and have also historically served as sources of conflict ... | [
{
"start": 0,
"end": 1820,
"label": 0
}
] | null | spans | null | proceedings | ACL | 2,019 | null | acl-2019-911d23b62de2a4e0 | paper:acl-2019-911d23b62de2a4e0 | null | null | 0 | 1,820 | 824 | 1,284 | null | null | stage1-epoch0-4270 |
82f2b308c3b01637c607ce05f52a2fed/g03/human_original | stage1-epoch0 | papers | paired | human | 0 | Artificial Neural Networks (ANNs), especially Deep Neural Networks, have become the go-to tool for many machine learning tasks. ANNs achieve state of the art performance in applications ranging from image classification and object recognition, to object tracking, signal processing, natural language processing, self dri... | [
{
"start": 0,
"end": 1441,
"label": 0
}
] | null | spans | null | proceedings | NeurIPS | 2,018 | null | 82f2b308c3b01637c607ce05f52a2fed | paper:82f2b308c3b01637c607ce05f52a2fed | null | null | 0 | 1,441 | 532 | 1,008 | null | null | stage1-epoch0-8013 |
acl-2019-f81eb6c15d85988b/g02/human_original | stage1-epoch0 | papers | paired | human | 0 | BERT (Devlin et al., 2018). Mean and max poolings are used as the latter two strategies. Results of various combinations of BERT type, layer and pooling strategy are illustrated in Table 8.
It’s reasonable that BERT trained on large model surpasses the smaller one in most cases due to the larger model capacity and ric... | [
{
"start": 0,
"end": 1863,
"label": 0
}
] | null | spans | null | proceedings | ACL | 2,019 | null | acl-2019-f81eb6c15d85988b | paper:acl-2019-f81eb6c15d85988b | null | null | 0 | 1,863 | 191 | 800 | null | null | stage1-epoch0-2551 |
luna-mirror/03c5d635398e30032d4ae1cdb17942ba15513f8e5b6b3182cb07cfe2b6843afc@0 | stage1-epoch0 | mirrors | generated | ai | 1,961 | licensed practitioner. It can be modified, transferred, or exchanged for a curse of comparable value.”
“I’d like no curse,” Mara said.
“Of course. No curse is an option, but it costs extra.”
“Why?”
“It’s a premium absence.”
Her brother leaned forward. “Can we negotiate the spoon curse? Maybe they only appear in s... | [
{
"start": 0,
"end": 1961,
"label": 1
}
] | null | spans | openai/gpt-6-luna | writingprompts | null | null | null | human:4fb06c1e86b06d13fc65e669ecb7fda2ff466c4b93415e4a0c905b1c90d9ddb2 | human:4fb06c1e86b06d13fc65e669ecb7fda2ff466c4b93415e4a0c905b1c90d9ddb2 | null | null | 872 | 2,833 | 0 | 1,961 | null | null | stage1-epoch0-1629 |
luna-mirror/ae831055687f4f56fb5169c814671de3c74167463f067e183ccd08505cda394c | stage1-epoch0 | mirrors | generated | ai | 2,288 | ordered us back from a passage that appeared perfectly safe to my untrained eye. Twice, after an hour’s detour, we found the channel there broken by rapids.
Near a settlement on a high red bank, we were invited to rest beneath a roof of palm thatch. The elder who received us, a woman named Nsimba, had traded palm oil... | [
{
"start": 0,
"end": 2288,
"label": 1
}
] | null | spans | openai/gpt-6-luna | gutenberg | null | null | null | human:f750cd2ad5b33031ad6999eb9fa57fd267f1f43bad1efcac023dde4bf9e79f4d | human:f750cd2ad5b33031ad6999eb9fa57fd267f1f43bad1efcac023dde4bf9e79f4d | null | null | 2,462 | 4,750 | 0 | 2,288 | null | null | stage1-epoch0-9086 |
183412de67059d8658622d232340b0cbae011d22a107f95763f74b70f86d80c6 | stage1-epoch0 | human | novel_human | human | 0 | New York Police Commissioner James O’Neill said around 3:05 p.m. local time, a man driving a rented commercial pickup truck entered the bike path, striking riders and pedestrians. The truck also struck a school bus, injuring two adults and two children.
The man then "exited the vehicle brandishing two handguns,” O’Nei... | [
{
"start": 0,
"end": 462,
"label": 0
}
] | null | spans | null | voa | null | null | null | human:bd57ac2c0e11e1e232fb617ea8307f2fe9cb5f7a2d0195d4b2cd1a8ff30ad773 | human:bd57ac2c0e11e1e232fb617ea8307f2fe9cb5f7a2d0195d4b2cd1a8ff30ad773 | null | null | 0 | 462 | 0 | 462 | null | null | stage1-epoch0-6943 |
luna-mirror/c82f85b4e48b7e3cefe997e8f3753c83c8a02342bc5045dbb428cb61454adf68@0 | stage1-epoch0 | mirrors | generated | ai | 2,816 | , initially because of the pandemic. Russia announced in 2023 that it was suspending participation. Moscow said it would continue observing the treaty’s central numerical limits, while Washington responded with measures intended to pressure Russia to return to compliance.
Neither side has announced a successor agreeme... | [
{
"start": 0,
"end": 2816,
"label": 1
}
] | null | spans | openai/gpt-6-luna | voa | null | null | null | human:38efad1c20a4c444053002d16e52643140606f8a53b4c5ed0a9fa156fd1cc521 | human:38efad1c20a4c444053002d16e52643140606f8a53b4c5ed0a9fa156fd1cc521 | null | null | 481 | 3,297 | 0 | 2,816 | null | null | stage1-epoch0-6066 |
acl-2016-39fec6314c5e012b/g02/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 678 | vs. 79 %), strict recall is not significantly different (bot at 67 %). Considering partial matches, G REEDY has significantly higher recall (81 % vs. 77 %) but significantly lower precision (88 % vs. 90 %) than the CRF, with an overall comparable F 1 . This result bolsters our hypothesis that the CRF learn only a smal... | [
{
"start": 0,
"end": 1011,
"label": 0
},
{
"start": 1011,
"end": 1689,
"label": 1
},
{
"start": 1689,
"end": 2296,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ACL | 2,016 | null | acl-2016-39fec6314c5e012b | paper:acl-2016-39fec6314c5e012b | null | null | 222 | 2,518 | 1,011 | 1,689 | null | null | stage1-epoch0-8753 |
fullpaper-261 | stage1-epoch0 | fullpapers | generated | ai | 1,999 | an exact diffusion reverse step. The student loss is
$$
\mathcal L_G=\mathbb E_{x,j}\!\left[\|\hat u-u_T\|_2^2+\kappa\|\hat u-u\|_1+\mu\|D_F(\hat u)-x\|_1+\nu d_f(D_F(\hat u),x)\right],
$$
where $D_F$ is the fixed pretrained image decoder and $d_f$ is a frozen feature distance. The teacher target helps preserve the ... | [
{
"start": 0,
"end": 1999,
"label": 1
}
] | null | spans | openai/gpt-6-sol | openreview | NeurIPS | 2,025 | null | neurips:uodE9CAXaF | paper:neurips:uodE9CAXaF | null | null | 6,924 | 8,923 | 0 | 1,999 | null | null | stage1-epoch0-4999 |
cffee3f2633293d48e890833586b046e0fbaa882ffe6279f988c8a7b28ecbdaa | stage1-epoch0 | human | novel_human | human | 0 | I don't know why people like The gospel of John better. I never had. Matthew is so much better. I saw an edited version of Matthew on tbn in '94 and thought it was great. and bought the 4 tape vhs in '96. and got the dvd in 2000. Matthew portrayel of Jesus was unlike any portrayel I ever saw before. from his tende... | [
{
"start": 0,
"end": 602,
"label": 0
}
] | null | spans | null | amazon2018 | null | null | null | human:882e40b1ce8081fbc649892910de91ed2a0a41a456fefaec50787b4caf92b6a9 | human:882e40b1ce8081fbc649892910de91ed2a0a41a456fefaec50787b4caf92b6a9 | null | null | 0 | 602 | 0 | 602 | null | null | stage1-epoch0-8202 |
44149063b0a3238b98db9485ead7ecc36393b857797ae699adadf9261d9d8e50 | stage1-epoch0 | human | novel_human | human | 0 | Meskel Square has some light relief from Addis Ababa after weeks of violence….some nice photos too of the hour. ………….The gloom that has settled over most people I know in Addis lifted for exactly one hour and 35 minutes over the weekend. The event was an official training morning for the upcoming Great Ethiopian Run, d... | [
{
"start": 0,
"end": 360,
"label": 0
}
] | null | spans | null | globalvoices | null | null | null | human:41f5e31ad438de4442ccb1ed88bfc4ff71e88b7172897de89cec94f61f9cdf6d | human:41f5e31ad438de4442ccb1ed88bfc4ff71e88b7172897de89cec94f61f9cdf6d | null | null | 0 | 360 | 0 | 360 | null | null | stage1-epoch0-9384 |
acl-2015-d7b892a22fcbce56/g04/human_original | stage1-epoch0 | papers | paired | human | 0 | Richard Sproat and Thomas Emerson. 2003. The first international chinese word segmentation bakeoff. In Proceedings of the second SIGHAN workshop on Chinese language processing-Volume 17, pages 133–143. Association for Computational Linguistics.
In this paper, we presented a series processes to reduce OOV rate and extr... | [
{
"start": 0,
"end": 1168,
"label": 0
}
] | null | spans | null | proceedings | ACL | 2,015 | null | acl-2015-d7b892a22fcbce56 | paper:acl-2015-d7b892a22fcbce56 | null | null | 0 | 1,168 | 246 | 936 | null | null | stage1-epoch0-8271 |
acl-2020-8915dc1fb602804d/g01/human_original | stage1-epoch0 | papers | paired | human | 0 | Utterance Insertion (UI) We remove each utterance of a dialogue and then re-insert it in any possible utterance position in the dialogue. We assume that the original place of the utterance is the best place for the insertion. Therefore, a coherence model ideally discriminates the original dialogue from the perturbed on... | [
{
"start": 0,
"end": 1403,
"label": 0
}
] | null | spans | null | proceedings | ACL | 2,020 | null | acl-2020-8915dc1fb602804d | paper:acl-2020-8915dc1fb602804d | null | null | 0 | 1,403 | 567 | 953 | null | null | stage1-epoch0-7697 |
acl-2018-0f9ac1c9fd29ac26/g03/paragraph_generate@0 | stage1-epoch0 | papers | paired | mixed | 581 | rases. We use 24,273 lexical paraphrases in their smallest set, PPDB-S (Pavlick et al., 2015), as equivalence relations, x y. The (partial) transformation function f for this generator is POS-tagged matched replacement of x in s with y, and the label g is entails.
SICK (Marelli et al., 2014) is a dataset of entailment... | [
{
"start": 0,
"end": 266,
"label": 0
},
{
"start": 266,
"end": 847,
"label": 1
},
{
"start": 847,
"end": 2140,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ACL | 2,018 | null | acl-2018-0f9ac1c9fd29ac26 | paper:acl-2018-0f9ac1c9fd29ac26 | null | null | 94 | 2,234 | 266 | 847 | null | null | stage1-epoch0-9612 |
acl-2017-ce6406e23cc51f62/g01/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 262 | The recall scores for all the gold spans are again averaged to get a global recall score. For the biomedical IE task, because we have gold labels for only a small set of 200 abstracts, we create 100 bootstrap re-samples of the (predicted and gold) spans and perform the evaluation for each re-sample. We then report the ... | [
{
"start": 0,
"end": 376,
"label": 0
},
{
"start": 376,
"end": 638,
"label": 1
},
{
"start": 638,
"end": 1055,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ACL | 2,017 | null | acl-2017-ce6406e23cc51f62 | paper:acl-2017-ce6406e23cc51f62 | null | null | 0 | 1,055 | 376 | 638 | null | null | stage1-epoch0-7672 |
a6299856495ad82bf717a2752a22a46e628a42080ec7e7ca15fd483c0e499675 | stage1-epoch0 | human | novel_human | human | 0 | If Lizardbeast rolls doubles he gets to move again, just like a human. If he rolls three doubles in a row, he does not move the third time, but neither does he go to jail. How is that supposed to work anyway? "Ladies and gentlemen of the jury, the facts are clear. Then he rolled two ones. Then, in the presence of nume... | [
{
"start": 0,
"end": 382,
"label": 0
}
] | null | spans | null | cccc | null | null | null | human:a9b3bf1cacb1bf7920b607eb14dc12b09776a8b95bef4216806052d6df8542f7 | human:a9b3bf1cacb1bf7920b607eb14dc12b09776a8b95bef4216806052d6df8542f7 | null | null | 0 | 382 | 0 | 382 | null | null | stage1-epoch0-8580 |
icml-2014-d3326281f94c2ffd/g03/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 418 | The planted clustering problems pose themselves as both statistical and computational problems. Statistically, the parameters n, r, K, p, q govern the “noisiness” of the problems: The problems become statistically harder with smaller values of p − q, K and larger r, as the observations are noisier and the cluster struc... | [
{
"start": 0,
"end": 493,
"label": 0
},
{
"start": 493,
"end": 911,
"label": 1
},
{
"start": 911,
"end": 1289,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ICML | 2,014 | null | icml-2014-d3326281f94c2ffd | paper:icml-2014-d3326281f94c2ffd | null | null | 0 | 1,289 | 493 | 911 | null | null | stage1-epoch0-2498 |
341283ca6780897b5f0bf0eb80b4b43c0db22c05e1d38019deceedbac632b1c8@40 | stage1-epoch0 | human | novel_human | human | 0 | 70--79 years189.9 Older than 80 years116.1
*Underlying disease*[a](#tblfn0005){ref-type="table-fn"} Cardiac147.7 Pulmonary5530.4 Others11965.7
*Comorbidities*[a](#tblfn0005){ref-type="table-fn"} Absence of comorbidities5932.6 Cardiovascular diseases189.9 Chronic diseases10759.1 Respiratory diseases2815.5 Others3720.4... | [
{
"start": 0,
"end": 1156,
"label": 0
}
] | null | spans | null | pmc | null | null | null | human:824a6a366c61b62f791d582aef043eb49e11919a138baec8a63e5e7f4320f1ec | human:824a6a366c61b62f791d582aef043eb49e11919a138baec8a63e5e7f4320f1ec | null | null | 623 | 1,779 | 0 | 1,156 | null | null | stage1-epoch0-924 |
luna-mirror/f7ca8ce0ec63f7d51c4c265324a64b3f1eb55987987311d1a2edbf9053f519e1 | stage1-epoch0 | mirrors | generated | ai | 828 | I bought this decorative tile set for a small kitchen backsplash and was pleasantly surprised by the range. The box includes geometric patterns, floral motifs, and several plain coordinating tiles, so it feels varied without looking chaotic. Colors are rich and consistent, while the printed detail looks crisp from norm... | [
{
"start": 0,
"end": 828,
"label": 1
}
] | null | spans | openai/gpt-6-luna | amazon2018 | null | null | null | human:6bba749f55c73aa36216efd9ccbe04ab2d131078b205fbf7cfc6b226f17c15aa | human:6bba749f55c73aa36216efd9ccbe04ab2d131078b205fbf7cfc6b226f17c15aa | null | null | 0 | 828 | 0 | 828 | null | null | stage1-epoch0-5488 |
luna-mirror/67b2383bdaf65fc8d9287047b3d607c899b5be054c732f01921c2c47f7ee695c | stage1-epoch0 | mirrors | generated | ai | 2,694 | an opportunity, not a guarantee, of peace. Taliban representatives described the withdrawal pledge as a victory, while also calling for the release of thousands of prisoners held by both sides.
President Donald Trump has said the United States wants to bring its troops home, but the agreement makes the withdrawal con... | [
{
"start": 0,
"end": 2694,
"label": 1
}
] | null | spans | openai/gpt-6-luna | voa | null | null | null | human:c21adb043ce53f3910d95ccb812ad9c8bb7bfeec2fefc08a66f17e2be94f2448 | human:c21adb043ce53f3910d95ccb812ad9c8bb7bfeec2fefc08a66f17e2be94f2448 | null | null | 1,285 | 3,979 | 0 | 2,694 | null | null | stage1-epoch0-5447 |
540089f217dfec79c7c1ad7d7ca34041452626a305b9c9b9f8669a7baaca498b | stage1-epoch0 | human | novel_human | human | 0 | Century Egg Or pidan (after the pine needle pattern which develops) is another Asian-egg delicacy, this time from China. Essentially a chicken, duck or quail egg is preserved and aged for a few weeks... or months... and then sliced and eaten with pickled ginger. Traditionally, the preservation process included packing... | [
{
"start": 0,
"end": 2301,
"label": 0
}
] | null | spans | null | foodista | null | null | null | human:06ba3fcf4df3908b55f0ddd31f39e2ee05483c46b2e5ecb5c782e0a00f55e5f2 | human:06ba3fcf4df3908b55f0ddd31f39e2ee05483c46b2e5ecb5c782e0a00f55e5f2 | null | null | 2,541 | 4,842 | 0 | 2,301 | null | null | stage1-epoch0-10824 |
luna-mirror/7346e1716038020be9666c2f14e97a6eb605c031bb251905cbe7757d8d02cec3 | stage1-epoch0 | mirrors | generated | ai | 1,503 | Public health researchers are testing new ways to forecast dengue outbreaks early enough for communities to act before hospitals fill.
The work combines several kinds of information, including rainfall, temperature, mosquito surveillance and reports of recent infections. Because dengue-carrying mosquitoes thrive in wa... | [
{
"start": 0,
"end": 1503,
"label": 1
}
] | null | spans | openai/gpt-6-luna | scidev | null | null | null | human:9f84c4cf1ba9b9a139a249c68a301d1e9a314528831bacdd6b37ff5159b164a7 | human:9f84c4cf1ba9b9a139a249c68a301d1e9a314528831bacdd6b37ff5159b164a7 | null | null | 0 | 1,503 | 0 | 1,503 | null | null | stage1-epoch0-8186 |
d65f082322734f450b75a482ea6978062728b3eb0fdcbfb1132b597a553fb0f8 | stage1-epoch0 | human | novel_human | human | 0 | Partitioning Integers using only elements from a specific set
I know how to partition given $p(x)$ using a generating function as my textbook on discrete mathematics explains it in detail. However, I want to know if it is possible to restrict the source elements of the partition? So, given I want to find the partition ... | [
{
"start": 0,
"end": 723,
"label": 0
}
] | null | spans | null | stack_tech | null | null | null | human:e1cb92d863d04218629dae7138eb0d5aa3ad6a2b3449c3d8f34768e7b452e831 | human:e1cb92d863d04218629dae7138eb0d5aa3ad6a2b3449c3d8f34768e7b452e831 | null | null | 0 | 723 | 0 | 723 | null | null | stage1-epoch0-4900 |
luna-mirror/0c356d5bd648193e916335f4e71d299aa97b824a1f7b3d1b00fafc57619eec15 | stage1-epoch0 | mirrors | generated | ai | 2,928 | Digital nystagmography (DNG) records eye movements using infrared cameras and provides quantitative measures that can complement clinical observation. In children referred for learning disorders, assessment may include spontaneous and gaze-evoked nystagmus, saccades, smooth pursuit, optokinetic responses, and, when ind... | [
{
"start": 0,
"end": 2928,
"label": 1
}
] | null | spans | openai/gpt-6-luna | pmc | null | null | null | human:3e7d472814bc027cef0ed68ccc9f4f658bc7561601cb581a49410ab12fe104b7 | human:3e7d472814bc027cef0ed68ccc9f4f658bc7561601cb581a49410ab12fe104b7 | null | null | 0 | 2,928 | 0 | 2,928 | null | null | stage1-epoch0-1628 |
a5e00132373a7031000fd987a3c9f87b/g01/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 555 | The goal of visual texture synthesis is to infer a generating process from an example texture, which then allows to produce arbitrarily many new samples of that texture. The evaluation criterion for the quality of the synthesised texture is usually human inspection and textures are successfully synthesised if a human o... | [
{
"start": 0,
"end": 386,
"label": 0
},
{
"start": 386,
"end": 941,
"label": 1
},
{
"start": 941,
"end": 2380,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | NeurIPS | 2,015 | null | a5e00132373a7031000fd987a3c9f87b | paper:a5e00132373a7031000fd987a3c9f87b | null | null | 0 | 2,380 | 386 | 941 | null | null | stage1-epoch0-9991 |
luna-mirror/bce5fcaa904f48b6f3314871a21f6262b9f0d82ea7b0a4ee87021bce7d14af30 | stage1-epoch0 | mirrors | generated | ai | 473 | Under the microscope, Mara watched Protympanium emerge from the sediment: a hollow central shell ringed by concentric lattices, with stout radial spines pointing like compass needles. She sketched its pores, then compared the specimen with nearby forms. Their shared skeletal geometry hinted at kinship among radiolarian... | [
{
"start": 0,
"end": 473,
"label": 1
}
] | null | spans | openai/gpt-6-luna | gutenberg | null | null | null | human:968ea01bb9688a07a6b7057d843858124cb8f3ac934548e2df5f873d99407c6c | human:968ea01bb9688a07a6b7057d843858124cb8f3ac934548e2df5f873d99407c6c | null | null | 0 | 473 | 0 | 473 | null | null | stage1-epoch0-10885 |
fullpaper-342 | stage1-epoch0 | fullpapers | generated | ai | 2,476 | a decision microblock the executable sequence is: freeze the policy and draw candidates; form and log contexts; sample arms from the recorded mixture probabilities; query the chosen arms and make policy updates; draw microblock probes before examining outcomes; run matched shadow branches for selected blocks; update a... | [
{
"start": 0,
"end": 2476,
"label": 1
}
] | null | spans | openai/gpt-6-sol | openreview | NeurIPS | 2,025 | null | neurips:tSpWkTFASC | paper:neurips:tSpWkTFASC | null | null | 12,935 | 15,411 | 0 | 2,476 | null | null | stage1-epoch0-9079 |
647bba344396e7c8170902bcf2e15551/g05/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 511 | Seriation seeks to reconstruct a linear order between variables using unsorted similarity information. It has direct applications in archeology and shotgun gene sequencing for example. We prove the equivalence between the seriation and the combinatorial 2-SUM problem (a quadratic minimization problem over permutations)... | [
{
"start": 0,
"end": 792,
"label": 0
},
{
"start": 792,
"end": 1303,
"label": 1
},
{
"start": 1303,
"end": 2163,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | NeurIPS | 2,013 | null | 647bba344396e7c8170902bcf2e15551 | paper:647bba344396e7c8170902bcf2e15551 | null | null | 0 | 2,163 | 792 | 1,303 | null | null | stage1-epoch0-3957 |
luna-mirror/b99f70ab148a11a62c28a22e91f76cdaf570352cbd108c707118977269fa0b51 | stage1-epoch0 | mirrors | generated | ai | 2,340 | This review considers how CASIS identifies, evaluates, and advances research for the International Space Station (ISS) National Laboratory, with particular attention to proposal pathways, flight implementation, and researcher experience. The process links scientific merit to practical feasibility: proposals must presen... | [
{
"start": 0,
"end": 2340,
"label": 1
}
] | null | spans | openai/gpt-6-luna | govreport | null | null | null | human:1082b7ee96ea9a527b43aa87aa15e1abfadd19a22ff426274e4c46b055a3d3fa | human:1082b7ee96ea9a527b43aa87aa15e1abfadd19a22ff426274e4c46b055a3d3fa | null | null | 0 | 2,340 | 0 | 2,340 | null | null | stage1-epoch0-8969 |
luna-mirror/e9f3d3f4dbe5f9c2f20750eab97e1dd7e030bc94a9ee752e4453a8ebce3e64a4 | stage1-epoch0 | mirrors | generated | ai | 2,942 | rule. But a country’s professed values are not proof that it honored them in practice.
The most useful accounts of the Cold War therefore combine ideology with agency. Communist doctrine shaped Soviet expectations about capitalism and class struggle; American faith in markets and liberal institutions shaped how U.S. ... | [
{
"start": 0,
"end": 2942,
"label": 1
}
] | null | spans | openai/gpt-6-luna | oanc_slate | null | null | null | human:a6f5a11125f881640ebce5122932bb485936a00dd53ced1c8713b99fa5d1c92f | human:a6f5a11125f881640ebce5122932bb485936a00dd53ced1c8713b99fa5d1c92f | null | null | 3,196 | 6,138 | 0 | 2,942 | null | null | stage1-epoch0-6245 |
acl-2017-51faf18f8c4a321c/g01/paragraph_generate | stage1-epoch0 | papers | paired | mixed | 424 | As mentioned in Section 3.7, this is also a challenge for human annotators because there may be several ways to decompose the sentence, many of which serve equally well to highlight certain malware aspects (see Figure 4).
Whether a decomposition is correct depends on the information that can be recovered from it. For ... | [
{
"start": 0,
"end": 223,
"label": 0
},
{
"start": 223,
"end": 647,
"label": 1
},
{
"start": 647,
"end": 966,
"label": 0
}
] | null | spans | openai/gpt-6-luna | proceedings | ACL | 2,017 | null | acl-2017-51faf18f8c4a321c | paper:acl-2017-51faf18f8c4a321c | null | null | 0 | 966 | 223 | 647 | null | null | stage1-epoch0-3635 |
luna-mirror/cc25f366bb566ec019e1edb9831a4ab35c2f1f94ec4a1b210ae7f215c92f252e | stage1-epoch0 | mirrors | generated | ai | 645 | I’d distinguish temporary crunch from the team’s normal operating model. In software, occasional releases run late, but chronic overtime often signals planning, staffing, or scope problems—not personal failure. Ask your manager what to deprioritize, document recurring after-hours demands, and protect recovery time. If ... | [
{
"start": 0,
"end": 645,
"label": 1
}
] | null | spans | openai/gpt-6-luna | stack_nontech | null | null | null | human:4ae79c7cfe9b441158489f1dcf37cac5be50be5fcdc68be65302de870ef77282 | human:4ae79c7cfe9b441158489f1dcf37cac5be50be5fcdc68be65302de870ef77282 | null | null | 0 | 645 | 0 | 645 | null | null | stage1-epoch0-5127 |
End of preview. Expand in Data Studio
Open Text Detector training mix: mixA
Not the mix behind the model on opentextdetector.com. That model was trained on
woog/open-text-detector-mix-oct7. mixA is a larger alternative (133,569 stage-2 rows) whose model scored within error of it on small edits; see the Evaluation page. That model was not released.
The training mix behind the mixA model (not released), a LoRA classifier on Qwen3.6-35B-A3B that labels each token, sentence and document of English text as human- or AI-written. The mix targets research papers, in particular small AI edits (one or a few sentences) inside otherwise human papers.
It holds 145,569 training rows: 12,000 in stage 1 and 133,569 in stage 2. mixA is built from the same pools as T2.1 with every AI pool used up to its screened supply and human text kept near T2.1's ratio. Stage 2 was seen once, with no repeated windows.
Splits
| Split | Rows | Use |
|---|---|---|
stage1 |
12,000 | Stage 1 (trains only the segment head). Drawn with replacement: 10,509 distinct windows. |
stage2 |
133,569 | Stage 2 (all heads and LoRA), one pass. |
selection |
2,100 | Windows from 195 held-out papers, used for the per-epoch selection loss. Not trained on. |
calibration |
2,019 | Windows from 195 further held-out papers, reserved for threshold calibration. Not trained on. |
selection and calibration are identical in the T2.1 and mixA releases, as is stage1.
Stage-2 composition
Labels: human = no AI characters, AI = no human characters, mixed = both (an AI span inside human context).
| Key | Rows | Human / mixed / AI | Writers | Description |
|---|---|---|---|---|
edit_claude |
13,332 | 0 / 13,332 / 0 | claude-opus-5-5 6,622, claude-sonnet-5-5 5,801, claude-haiku-4-5 909 | One to a few sentences inserted or rewritten by Claude inside a human paper passage |
edit_luna |
12,432 | 0 / 12,432 / 0 | gpt-6-luna 12,432 | Sentence-level edits and splices by GPT-6 Luna inside human paper passages |
edit_deepseek |
3,342 | 0 / 3,342 / 0 | deepseek-v4.1-flash 3,342 | Sentence-level edits by DeepSeek V4.1 Flash inside human paper passages |
edit_gpt |
378 | 0 / 378 / 0 | gpt-6.1-sol 378 | Sentence-level edits by GPT-6.1 Sol inside human paper passages |
para_luna |
4,596 | 0 / 4,593 / 3 | gpt-6-luna 4,596 | A paragraph regenerated by GPT-6 Luna, in its original human context |
para_e |
3,528 | 0 / 3,303 / 225 | claude-sonnet-5-5 1,810, claude-opus-5-5 1,330, claude-haiku-4-5 217, gpt-6.1-sol 171 | A paragraph rewritten or drafted by Claude or GPT-6.1 Sol, in human context |
para_deepseek |
810 | 0 / 810 / 0 | deepseek-v4.1-flash 810 | A paragraph drafted by DeepSeek V4.1 Flash, in human context |
sections_ai |
4,320 | 0 / 1,014 / 3,306 | claude-opus-5-5 2,011, claude-sonnet-5-5 1,999, claude-haiku-4-5 310 | A paper section written by Claude from the paper's surrounding text |
mirrors |
27,759 | 0 / 0 / 27,759 | gpt-6-luna 27,759 | Fully AI documents written by GPT-6 Luna to mirror a human passage's genre and topic |
fullpapers |
4,122 | 0 / 0 / 4,122 | gpt-6.1-sol 2,140, gpt-6-luna 1,982 | Windows from full manuscripts written by GPT models from a human title and abstract (body only) |
claude_fp |
5,550 | 0 / 0 / 5,550 | claude-opus-5-5 2,498, claude-sonnet-5-5 2,300, claude-haiku-4-5 752 | Windows from full manuscripts written by Claude from a human title and abstract (body only) |
human_paper_orig |
24,331 | 24,331 / 0 / 0 | — | Unedited human paper text (paired originals and clean-v2 extractions) |
human_section |
4,306 | 4,306 / 0 / 0 | — | Human originals of the sections in sections_ai |
human_para_e |
4,341 | 4,341 / 0 / 0 | — | Human originals matched to para_e |
human_host |
3,022 | 3,022 / 0 / 0 | — | Unedited human host passages of the kind used for edits |
human |
10,200 | 10,200 / 0 / 0 | — | Generic human text from 34 non-paper and paper sources |
gradtex |
7,200 | 3,539 / 0 / 3,661 | — | GRADTEX (derived from MAGE); document-level labels only |
Stage 1: papers 5,400, mirrors 3,000, human 3,000, fullpapers 600. Its papers rows are paired NeurIPS/ICML/ACL passages; fullpapers are GPT-6 Luna, GPT-6 Sol and GPT-6.1 Sol manuscripts.
AI writers, all training rows: openai/gpt-6-luna 52,699, anthropic/claude-opus-5-5 12,461, anthropic/claude-sonnet-5-5 11,910, deepseek/deepseek-v4.1-flash 4,152, openai/gpt-6.1-sol 2,886, anthropic/claude-haiku-4-5 2,188, openai/gpt-6-sol 207. GRADTEX AI rows come from the generators of its upstream sources and carry no per-row writer. No Claude Fable output is included; it is reserved for evaluation.
Small edits. In stage-2 mixed rows, the AI span has a median of 202 characters (5th–95th percentile 75–1127); 19,347 rows have under 200 AI characters.
Sources of the human text
- Research papers, all dated 2022 or earlier: NeurIPS, ICML and ACL proceedings, 2013–2021 (
source=proceedings); OpenReview papers from ICLR, NeurIPS, CoRL and TMLR (openreview); and open-access passages from PMC, arXiv, peS2o and the ACL Anthology (pmc,arxiv,pes2o,acl).venueandyearare filled where known. - Generic human text (
human, and the seeds ofmirrors): gutenberg 4,694, writingprompts 3,413, wikipedia 3,230, pmc 2,412, stack_nontech 2,126, arxiv 2,118, cccc 1,879, amazon2018 1,864, pes2o 1,594, pressbooks 1,471, acl 1,307, opinrank 1,110, libretexts 957, voa 782, stack_tech 777, globalvoices 726, wikibooks 692, foodista 682, scp 665, ubuntu_irc 626, oanc_slate 583, govreport 558, wikisource 553, wikivoyage 524, imdb 427, wikinews 425, wiki_talk 405, scidev 352, worldbank 342, fed 318, eff 192, oanc_icic 155. - Full-manuscript seeds: the human title and abstract of an existing paper. Only the generated body is included, so no seed text appears.
fullpapersseeds are ICLR 2026, NeurIPS 2025, ICML 2025 and COLM 2025 papers;claude_fpseeds are half from 2022 or earlier and half from those 2025–2026 venues.venueandyeargive the seed paper. - GRADTEX, as published upstream.
Construction
- Long documents were cut into non-overlapping windows that fit 510 tokens under the Qwen3.5, Qwen3.6 and Gemma tokenizers. For paired and edited rows the window is placed around the AI span.
- Stage-2 rows were screened against every held-out evaluation list: rows from evaluation, selection or calibration papers, or sharing a sentence of 10+ words with them, were dropped. Stage 1 is copied unchanged from the earlier T2.1 preparation.
- Allocation per key was round-robin over papers. When a key ran short it was filled from another key in the same group (for example
edit_gpthad only 378 usable rows).
Fields
| Field | Meaning |
|---|---|
id |
Row id; encodes the source document, edit and window offset. Unique in stage2, selection and calibration. |
split_file, draw_id |
Original training file and draw position. |
dataset, kind |
Mix key (table above) and row type. |
label |
human, mixed or ai, derived from regions (or document_label). |
text |
The window. |
regions |
Character spans of text with label 1 = AI, 0 = human. The source of the token, sentence and segment labels. |
document_label, supervision |
GRADTEX rows only: document_only supervision with a document label (1 = AI). Their regions are empty. |
ai_chars |
Number of AI-labelled characters. |
generator |
Writer of the AI text; null for human-only rows and GRADTEX. |
source, venue, year |
Origin of the human text or seed (see above). |
text_format |
Text extraction lineage: paired_pdf_v1, clean_v2, archive_v1 (paper PDF extractions), host_* (open-access host passages), generic_human, generated_doc, generated_markdown, gradtex. |
paper_id, group, host_id, seed_id |
Source identifiers. Split by group to keep a document's windows together. |
source_start, source_end, target_start, target_end |
Window offsets in the source document, and the span around the edit within the window. |
original_split, protected |
Bookkeeping from the upstream paired-paper split. |
Known limitations
- Text format correlates with the label. In stage 2, the share of non-human rows is 0.26 for
clean_v2(32,753 rows); 0.78 forpaired_pdf_v1(28,575 rows); 0.66 forhost_pmc(8,209 rows); 0.74 forhost_arxiv(3,996 rows); 0.69 forhost_pes2o(4,464 rows). A model can partly learn format as a shortcut. - GRADTEX labels are document-level only.
stage2contains 343 rows whose text duplicates another row.
Rights
This collection combines material under different terms and grants no new rights to any part of it.
- Human text keeps the terms of its source: per-paper terms for conference papers, per-article open-access licenses for PMC, arXiv, peS2o and ACL Anthology text, CC BY-SA for Wikimedia projects, and the release terms of each research corpus (including GRADTEX).
- AI-written text is subject to the terms of the providers of the listed models.
Please open a discussion to request removal of any item.
- Downloads last month
- 87